Untitled

HOME
SHOSHONE NEWS-PRESS, January 29, 2002, pp. 6 & 10.

Common Sense

Agency's blood lead model lacks foundation for unverified trust in basin
By Ron Roizen, Ph.D.
SNRC Science Committee

In the previous column I argued EPA accorded its IEUBK computer simulation model for estimating childhood blood-lead levels an aloof, "geo-platonic" status protected from actual empirical testing. I also suggested that though there were credible reasons for this philosophical disposition, nevertheless the EPA's aversion to empirical checking could harbor and protect faulty science. Finally, I suggested the IEUBK model, as it has been applied in the basin, does not merit the trust that the EPA has vested in it.
Why my distrustfulness? The EPA developed not one but instead multiple sets of IEUBK-based predictions or estimates of childhood blood-lead levels in basin communities using different input assumptions in the model. Two of these model variations -- the "Default Model" and the "Box Model" -- preoccupied much of the analysis offered in the Human Health Risk Assessment report. A fundamental scientific weakness is evidenced by the fact that these alternative IEUBK modes -- DM and BM -- generated very different blood-level estimates or predictions for basin communities.
For example, for Mullan the DM predicted that just under half (48%) of children under 8 years old had blood-lead levels over 10 µg/dL whereas the BM predicted fewer than one child in five (19%) exceeded that same standard. In Kingston, DM predicted 11% of children exceeded the standard wheras BM predicted only 2% did. Similar large disparities characterized the predictions for all basin communites (see chart, page 10).
As in any predictive situation -- whether an auto mechanic's estimate of potential repair costs or an astronomer's prediction of where a planet will be on a given date -- very different predictions suggest a relatively low state of knowledge. Very different predictions also erode whatever scientific authority the IEUBK model may possess because we, the consumer of this science, are left not knowing which prediction should get the benefit of that scientific authority.
Region X might have taken the average of the two disparate estimates -- using the original estimates to define a margin of error around that mean. (In fact, EPA's Technical Review Workgroup for Lead, which reviewed an earlier draft of the Human Health Risk Assessment study, suggested something like this tactic.) Yet that approach could not fully undo the confidence- and authority-eroding impacts of the wide disparities in the original estimates.
Alternative "bioavailability" assumptions account for the largest part of the difference between the DM and BM variations of the model. "Bioavailability" refers to the fact that different lead compounds or "species" have different capacities for converting lead exposure into bloodstream lead. The BM version of the IEUBK assumes a bioavailability factor that is 40% lower than the DM version (i.e., 18% bioavailability in the BM and 30% bioavailability in the DM).
The authors of the basin HHRA decided that the BM's predictions were superior to the DM's predictions for the upper basin area. They employed the BM variation of the IEUBK for determining soil remediation goals.
The HHRA's main text creates the impression that blood-lead survey gathered between 1996 and 1999 held a significant place in the decision to choose the BM over the DM predictions. (These same survey data supply the "observed" bar in the chart borrowed from the HHRA report.) The science committee vigorously objected to this decision procedure, given that these survey data derived from nonrepresentative samples and involved multiple counts of the same individuals.
Appendix Q of the HHRA, however, makes it clear that the main case for the BM variation of the IEUBK model derived from the predictive experience of the IEUBK inside the Bunker Hill Superfund Site Box. The regular-issue version of the IEUBK "overpredicted" blood-lead levels over an 8-year period in the history of the BHSS. The IEUBK's DM version predicted significantly higher blood-lead exceedance rates than blood-lead surveys inside the Box reported. Incidentally, this tendency to overpredict exceedance rates declined with time -- thus creating a fascinating analytical puzzle for the study's scientists. Adjustments made to bring the IEUBK's predictions into closer agreement with blood-lead survey measures inside the Box provided the chief scientific reason for selecting the BM version over the DM version of the IEUBK in the wider basin area.
But applying Box-based adjustments to the IEUBK for use in the wider basin raises numerous and important scientific problems. Lead-speciation in the wider basin may be quite different from that inside the Box. Moreover, the inside-the-box blood-lead study did not employ a control-group or control-site design, which design would have allowed researchers to assess how much of the inside-the-Box blood-lead decline was due to Superfund remediations and how much to wider trends affecting children outside the Box as well as inside. No actual lead speciation was carried out inside (or outside) the Box -- a fact which both renders the lead-exposure science incomplete and deviates from official EPA policy regarding the IEUBK's use and modification. Finally, the inside-the-Box historical analysis was essentially a post hoc curve-fitting enterprise, which means in turn that the application of the inside-the-Box BM version of the IEUBK to the wider basin may be regarded as an hypothesis-testing exercise and not as confirmatory science.
The EPA's Technical Review Workgroup for Lead (TRW) was sensitive to the HHRA's weaknesses and clearly uncomfortable with HHRA's choice to go with the BM version of the IEUBK for the basin.
Unfortunately, the TRW's considered range of bioavailability options extended only from the BM's 18% to the Default Model's 30%. Though this fact may express EPA's conservative policy regarding when modifications in the IEUBK's bioavailability parameter may be permitted, the science committee has repeatedly noted that there is no "law of nature" nor other barrier that disallows still lower true bioavailability levels in basin soils. Indeed, a recently published study of variability in relative bioavailability of lead revealed a range of 3% to 86% bioavailability across soil samples (Casteel et al., Int J Hyg Environ Health 203:473-4, 2001).
It bears noting that the above account of scientific incompleteness and the resulting uncompelling character of the HHRA's chief conclusions apply to a site that
it is said would become the largest Superfund in the history of EPA. That fact alone should oblige Region X to conduct a full and unambiguous study of lead speciation/bioavailability and childhood blood-lead levels in the basin before any remediation enterprise is contemplated, designed, or put into effect.

[SIDEBAR]
Quotes illustrate concerns
Some selected quotations from the TRW's review are offered in this sidebar. These serve to illustrate the TRW's concerns and unease with the HHRA's scientific choices -- though the reader is cautioned that spot quotation always risks "quotation out of context.")
Spot quotations from the TRW's review of the draft Human Health Risk Assessment for the CDA basin:
"The Box model was calibrated to agree with the downward trend in post-remediation blood lead concentrations observed at the BHSS. Factors that may have affected this downward trend...may not be operating or may not be as important in the CDAB. Ideally, if adjustments were to be made to the IEUBK model for its application to the CDAB, such adjustments should be based on the available information about exposures and blood lead concentrations in the CDAB and not at the BHSS" (Appendix W, p. 120).
"At this time, there does not appear to be an adequate basis for determining which of the two models
provide more accurate risk predictions in the CDAB" (App. W, p. 120).
"Another uncertainty in extrapolating a bioavailability factor for the CDAB from BHSS data is that it is possible that exposures in the CDAB may be a mix of lead from the smelter and lead from mine wastes, or other sources, which may have different absorption fractions. The relative contribution of these sources may change with location in the CDAB (e.g., with upwind or downwind from the smelter, or up or down gradient from the smelter), and may change with remediation. For example, at some locations in the CDAB, historic smelter emissions may contribute more to lead in house dust than in yard soils. If lead in smelter dust has a different fractional absorption than lead from other sources, removal of yard soil may change the absorption fraction of the lead to which children would be exposed at that location. There is some support for this possibility in the BHSS, where the calculated bioavailability factor which resulted in better agreement between the IEUBK model predictions and observed blood lead concentrations changed (increased) over time as the remediation proceeded (see Appendix Q, HHRA)" (App. W, pp. 121-122).

"Common Sense" is a column by members of the Shoshone Natural Resouces Coalition's Science Committee. Responses are invited from those who agree and those who disagree.