Peer Review of the Surface Impoundment Study (SIS) Technical Plan for Human Health and Ecological Risk AssessmentGeneral Comments and Recommendations by Naida Gravelis
My review of the Surface Impoundment Study Technical Plan for Human Health and Ecological Risk Assessment (the SIS Technical Plan) focused on those sections pertinent to human health risk. General comments are summarized below. Page-specific comments and comments related to the specific topics for which EPA solicited peer reviewer input follow the general comments.
Overall approach: The overall approach and methodologies presented in the SIS Technical Plan are generally scientifically defensible. However, the presentation of proposed "screening" methodologies is at times confusing and internally inconsistent. The approach is not always laid out in a logical sequential pattern. The examples help. This is particularly true for the Phase IA discussions.
Distinguishing between "screening criteria" and "risk criteria": The "dual objectives" of the risk analysis (i.e., screening out constituents/units with negligible risk and building risk profiles) results in some of the confusion mentioned above. The plan does not consistently describe how/when the screening versus the risk decision criteria will be applied. The plan should more explicitly state under what circumstances a constituent, etc. will be removed from further consideration. The plan should also make it more clear that all constituents will be carried through when generating risk distributions. The executive summary (e.g., Page 1-7, ¶1 and Page 1-10, ¶2) does a better job in communicating the components of the process in a step-wise fashion than does Chapter 2. (See also Question #3 below.)
Level of protectiveness: Further justification is needed for some of the proposed exposure criteria used in the initial screening analysis. For example, use of central tendency values in the Phase IA screen warrants further discussion. Also, the criteria for applying a "margin of protection" (MP) needs to be more definitively described. (See also Question #3 below.)
Defining terms: It would be helpful if certain terms were defined (e.g., in-place receptors, in-place closure scenarios).
Feasibility of the Phase II Effort: Depending on the outcome of the Phase I screen (i.e., the number of constituents, impoundments, and facilities requiring further analysis), the Phase II assessment may be overambitious, but this is acknowledged in the plan. The proposed ranking system(s) (Phase IC) and EPA's proposed alternatives for Phase II modeling (i.e., representative versus site-specific) seems reasonable, both of which are aimed at prioritizing and simplifying.
The representativeness and quality of the survey data: The quality and reliability of the risk assessment results are dependent on the representativeness and quality of the survey data, both in terms of the facilities selected for analysis and the data they provide. The random selection of the direct and zero dischargers from the survey appears to represent a cross-section of industry types (although it would be helpful to see a similar breakdown for the selected facilities as was done for the "study-eligible" facilities in Figure 1-2). EPA also took into consideration possible bias due to nonresponse. It is less clear whether the "purposive" sample of 35 indirect discharge facilities (as described in Appendix B) is representative.
The representativeness and quality of the measured constituent data also are critical to the reliability of the risk analysis. Therefore, EPA needs to clearly demonstrate that (1) the available data meet the criteria for data usability for risk assessment, and (2) the data are reasonably representative of conditions over time and across facility types. Also, is there a mechanism in place to verify the accuracy of the other survey information that will be incorporated into the Phase II risk analysis (e.g., receptor locations; soil and hydrogeologic conditions)?
Inclusion of the survey: Although the survey tool is readily available on-line, it would be helpful to have a copy of the survey included within the SIS Technical Plan for reference.
Response to Charge Questions
#1 Derivation of human health screening factors
Is the methodology presented for calculating screening factors suitable?
In general, the overall approach and underlying principles for deriving human health screening factors are scientifically sound. That is, the formulas and methodologies (Table 2-1 equations) are consistent with generally accepted risk assessment practices. (Issues related to the appropriateness of the proposed exposure parameters used in these equations are discussed in Question #3 below.)
As mentioned in the general comments, the procedures related to the application of the screening factors in the Phase IA human health screening is confusing at times. The concept of using the screening factors to simplify the "standard forward calculation of risk" should be emphasized. It is, therefore, recommended that Section 2.2 be reworked to first clearly present the methodology for developing the screening factors and then introduce the "risk equation" (page 2-4) that will be used to generate the initial risk distributions. In addition, it would have been helpful to introduce the concepts presented in Figures 2-3 and 2-4 earlier in the Phase IA discussions. Otherwise, the screening concept gets masked and the exercise at first glance seems to be exclusively a forward risk assessment exercise.
Are there readily available data sources that would provide additional benchmark values?
The SIS Technical Plan identifies the most reliable benchmark values (i.e., those listed in IRIS and HEAST, interim values derived by NCEA, as well as toxicity benchmark values derived by ATSDR and CalEPA). When discussing toxicity benchmarks as part of the Phase IA screening (Section 18.104.22.168, page 2-6), EPA also proposes using the provisional approach developed by HWIR to develop interim toxicity benchmark[s] for screening purposes. This methodology should be referenced.
Section 22.214.171.124 (Human Health Effect Benchmark) presents the typical hierarchy for choosing toxicity values for use in risk assessment and is therefore appropriate. The proposed benchmarks (page 3-24) for several of the 25 chemicals without benchmark values are reasonable based on chemical characteristics and erring on the side of conservatism in each case. For fluoride, using the RfD for fluorine (0.06 mg/kg/day) is valid; it is consistent with ATSDR's MRL of 0.05 mg/kg/day for sodium fluoride (which does not appear in Appendix A).
#2 Derivation of ecological screening factors
As is the case with the human health screening factors, the proposed derivation of ecological screening factors appears to be logical. Typically used benchmark values are presented. I defer, however, to the peer reviewers with expertise in ecological risk assessment for a more comprehensive assessment of the ecological screening factors.
#3 Level of protectiveness
Is the screening process designed in such a way to enable the Agency to identify constituents, impoundments, and facilities with negligible risk?
Two issues need further attention or clarification to fully demonstrate the level of protectiveness that the plan offers. First, will an MP of 0.1 be used globally in the Phase IA screen? It should be. Second, further justification is needed for using mean concentrations and central-tendency exposure parameters in the initial screening analysis.
Using the MP factor: It is implied, but not consistently stated that an MP of 0.1 will be applied to the proposed risk criteria, resulting in the use of a 10-6 screen for carcinogens and a hazard index (HI) of 0.1. Providing this added level of protection during Phase I is appropriate-to be adequately protective and to ensure that chemicals with additive effects or indirect exposures are not prematurely eliminated. The executive summary (page 1-8) states that EPA "will" apply this added margin of protection in the initial screening. Chapter 2 is less explicit. On page 2-3 (last line), page 2-20 (¶2), and page 2-22 (¶1), it is indicated that EPA "may" choose to use an MP. The sample bar graphs showing risk distributions also do not communicate a consistent message. That is, when will constituents/units require no further analysis? For example, Figure 1-3 (page 1-10) provides a sample of Phase IA screening results, noting that constituents above the "screening criteria" will proceed to Phase IB. The "screening criteria" in this case include an MP of 0.1. Figure 2-11 (page 2-26), on the other hand, indicates that only those constituents exceeding the risk criteria (not the screening criteria) will proceed to Phase IB.
Using mean concentrations and central-tendency exposure assumptions: High-end or "worst case" assumptions are generally more acceptable for screening evaluations. The justification offered in the plan for using mean concentrations in the screening analysis is limited to (1) the belief that it is highly unlikely that receptors would come in contact with the constituent concentration measured in these media or be exposed to maximum concentrations over time, and (2) "highly protective exposure assumptions" are used in the risk analysis (page 2-13, ¶4). The average concentrations will come from the survey and the results of EPA sampling. Understanding the representativeness of these data sets is key to judging whether using average concentrations is protective enough. The questionnaire provided on line asks for sampling date, number of samples, and averaging period (for data collected over the past 3 calendar years), but the SIS Technical Plan does not specify how the quality and representativeness of these data will be assessed. The survey also asks for reasons for variability in a facility's data (e.g., production cycles, seasonal influences). Again, how will the significance of these responses be judged. How will the data quality be measured/verified? Regarding the second point, the proposed exposure assumptions represent "typical or central-tendency" values, not high-end values (page 2-9, ¶1) and are, therefore, not necessarily "highly" protective.
With this said, the question remains whether the proposed screening process truly meet the objective of screening out constituents and units with "negligible" risk. Although I agree that it is highly unlikely that receptors will contact surface impoundment contaminants at the source, using the mean concentration data coupled with central-tendency exposure assumptions could potentially screen out more than "negligible" risk constituents. At a minimum, the MP of 0.1 should be applied in all cases.
Subsequent phases of the screening and risk assessment process (Phase IB and Phase II) appear to have adequate conservatism built in, assuming no shortcomings are identified in the models. These tools will enable EPA to refine the Phase IA risk estimates.
Special case constituents (Section 2.2.3, page 2-36): Some clarification is needed in terms of how/when EPA's Revised Waste Minimization Prioritization Tool (WMPT) will be worked into the screening analysis to identify toxic constituents that may persist or bioaccumulate, but may have otherwise been screened out of the initial screening analysis. How will these constituents be handled? Will high-scoring constituents be automatically carried to Phase II? Will the screening risk values generated for such constituents be considered?
#4 Approach for dealing with lack of information on chemical composition of wastewater in the impoundments or emissions
The use of data from other facilities requires knowledge or making the assumption that the 221 facilities to be used in this analysis are representative, both in terms of wastewater composition/concentration and emissions data quality. Assuming this is true, the approaches laid out in the technical plan are reasonable. The decision tree for selecting surrogate water data when no survey data are available (Figure 2-7) is logical. Although somewhat implied, the plan should be more explicit about the criteria/hierarchy for selecting surrogate air and sludge data. The discussions in the plan related to estimating concentrations/choosing surrogates are limited to one paragraph (page 2-13, ¶3).
#5 Approach for representing cumulative risks
Within the scope of the SIS Technical Plan, EPA's proposed approach for representing cumulative risks is sound. The approach is consistent with generally acceptable practices in risk assessment. In generating risk profiles, caution must be taken when generating aggregate risks from inhalation and ingestion exposures. Because assessment of the inhalation pathway considers both portal-of-entry and extrarespiratory effects, it is important that target organ considerations be carefully examined within and across exposure routes. This information is not presented in the summary tables within the technical plan for verification, but the importance of this issue is flagged within Phase II discussions (e.g., Page 3-26).
The Phase II modeling effort will generate multi-pathway, multi-chemical, and multi-receptor risks. The identified receptor groups and the proposed combinations of exposure pathways presented in Chapter 3 and Appendix D seem reasonable.
#6 Modeling approaches
The proposed models (i.e., IWAIR, IWEM, and 3MRA) are undergoing review, have undergone independent peer review, or are in the public comment period. The outcome of these reviews is critical to verifying the utility of these models in this screening and risk assessment process. Of particular interest will be issues related to model validation and uncertainty analysis.
For the purposes of this peer review of the SIS Technical Plan, the proposed screening models (IWAIR and IWEM) were evaluated for their ability to produce conservative estimates of air and water exposures. For the Phase II analysis, the 3MRA model was reviewed and evaluated for whether it will (1) produce reasonable exposure point concentrations, (2) incorporate reasonably protective exposure assumptions, (3) use appropriate human health benchmarks, and (4) generate realistic predictions of human health risk.
IWAIR/IWEM: The use of these models are appropriate for their intended use. Both models yield conservative predictions of possible exposure point concentrations.
3MRA: While I do not have the expertise to fully evaluate the functionality and utility of this type of multi-faceted predictive model, the modeling capabilities offered by the 3MRA modules as described in the plan appear to provide for a comprehensive evaluation of multi-pathway/multi-media risks. The assumptions for the individual modules are reasonable. In addition, accounting for source contribution (direct and indirect), multiple fate and transport processes, and uptake in the foodchain promises to provide realistic predictions of exposure point concentrations. Will any field testing of selected facilities be performed to help validate the model predictions?
Certain shortcomings/limitations exist, however. EPA recognizes two important limitations (i.e., the inability of 3MRA to model multiple impoundments at a site and to model a postclosure scenario). The measures that EPA proposes to take for assessing multiple surface impoundments (Section 126.96.36.199) are reasonable, but the data collection needs and postprocessing techniques for the Phase II risk results need to be clearly laid out in the plan. The proposed modifications to the land application unit module appear to reasonably address the postclosure scenario. .Miscellaneous Page-specific Comments
Page 1-5, ¶1: Additional information regarding the 1997-1998 pilot study would enable a more informed assessment regarding the representativeness of the survey data.
Page 1-1: Section 188.8.131.52 introduces EPA's proposal to develop representative hydrogeologic and water shed scenarios in the event Phase I produces a large number of sites requiring Phase II analysis. This section also notes that the Agency is considering this same type of approach to include representative ranges of populations exposed; however, the plan does not elaborate here or later on what this might involve.
Page 1-12: Section 184.108.40.206 indicates that high-end and central tendency scenarios (based on real receptors) will provide a realistic span of potential risks in the Phase II analysis. While the model is presumably capable of generating risk distributions for a range of values, subsequent text and Appendix D only include brief reference to the production of high-end and central tendency model outputs. Appendix D presentations of exposure assumptions (e.g., exposure duration = 9 years), where provided, appear to be central-tendency assumptions.
Page 2-9: Table 2-2 (Exposure Parameter Values): As mentioned above, selection of central tendency values for a screening level risk analysis is questionable mostly from a perception point of view. Most notable are inhalation and water ingestion rates. Using 90th percentile values would more clearly demonstrate that only constituents/units with negligible risks are being removed from the analysis.
Page 2-12: Figure 2-4 does not show when one should proceed to Phase IB for the groundwater pathway.
Page 2-30: Typographical error: Top left box in Figure 2-12 should read "see Figures 2-13 and 2-14" and the top right box should read "see Figure 2-15."
Page 2-31: Cumulative Risk Calculation: The plan indicates that Phase IA and Phase IB risk estimates are "combined." For clarity, it is recommended that the plan be more explicit that Phase IB results refine the Phase IA risk estimates by industry (as shown in Figure 2-10). Specifically, the Phase IB results end up replacing those estimates previously calculated for water. Figure 2-16 helps to illustrate this.
Page 2-35: For the combined Phase IA and IB analysis shown in Figure 2-16, why doesn't the sludge ingestion risk for Phase IA show?
Page 2-64: Table 2-11: Text discussions on page 2-63 suggest that "human health individual excess cancer risk bin" for the cancer risks "> 10-6 and < 10-5" would not be carried through to Phase II, yet in Table 2 this bin is given a score of "3." This requires clarification.
Page 3-18: What is "U.S. EPA 1997d, in press?"
Page 3-19: Section 220.127.116.11 indicates that exposure via groundwater will be evaluated for private groundwater wells only. Will potentially affected public supplies be addressed in any way?
Page 3-22: When toxicity benchmarks are not available, the plan indicates that interim benchmarks will be developed from primary scientific literature. Will this be done through NCEA or EPA? The basis for selecting applicable literature values and the criteria used to develop interim values should be described. The plan appropriately indicates that analytic findings based on draft/interim benchmarks will be clearly identified.
Page 3-27: "Individual vs. Population-Weighted Risk Estimates" and "Uncertainty Analysis": The text indicates that estimates of risk will be described in terms of "the number of individuals who...can reasonably be expected to incur risks from constituents from impoundments..." It should be noted that these risk predictions are based on numerous assumptions and the analysis includes limitations and uncertainties that will need to be carefully communicated. Therefore, the uncertainty analysis should evaluate the variability as indicated, but also indicate how different factors may affect risks estimates, to the extent possible (e.g., over or under estimate risks).
Page 3-46: Section 3.3: The number of simulation modules based on Figure 3-13 and subsequent text descriptions is 13, not 14.
Appendix D Do the outputs from the 3MRA modules provide central tendency and high-end values? No comments.