The Human Health Risk Assessment for both marsh sediments and upland soils tries to measure the danger to people who use the LCP site. Based on the findings in the report, there appears to be some risk of toxic exposure to present and future site workers, future residents, and anyone eating fish or shellfish from the marsh waters.

The study found non-cancer risks were high for future residents of the site and anyone ingesting seafood caught in the area. Increased cancer risks were not found.

For site workers the Hazard index equaled 1; the EPA generally regards a value in excess of 1 as hazardous, in this case, technically, the value does not exceed 1. However, in building the model for the LCP site, the study uses value averaging and exposure assumptions that are not true for all portions of the site. In examining the actual site data, high concentrations of PCB occurs in very specific locations. Since PCB is the major chemical driving the excess risk, these PCB site locations are hazardous, not borderline as indicated in the Risk Assessment.

Where the model is particularly deficient is at the marsh shoreline where there is only minimal cleanup. The toxin values for chemicals such as mercury and PCB are quite high, and direct contact with marsh soil and water can occur with greater frequency than predicted. Other aspects where the model may underestimate long-term risk include the "historic" buildings on-site and the risk from contaminated seafood.

The community should regard the elevated risk remaining at the site as unacceptable for immediate use, both commercially and residentially. Worker safety cannot be assured at the site until some areas are either further cleaned, or all access to those contaminated areas is positively eliminated.

The Glynn Environmental Coalition (GEC) received for review the Human Health Baseline Risk Assessment, Marsh Sediment and Upland Soil, LCP Chemicals Site final report (October, 1997) along with the draft of this document, Environmental Protection Agency (EPA) responses to draft comments, and a cover letter from the Potentially Responsible Parties (PRPs) arguing several points regarding EPA policy. In addition, the GEC received the Cultural Resources Literature and Records Review of the LCP Chemicals Site and Former Arco Refinery, Brunswick, Glynn County Georgia (October, 1997). The cultural review contains information that can impact the long-term use of the site, thus affecting the exposure model; consequently, these documents were reviewed together.

This study examines the sites’ use by people from prehistoric times to the present. The purpose is to determine if the site contains any cultural artifacts that are of interest to present or future generations of scientists, historians, and residents.

Overall, the study found that the intense industrial activity over the last 200 years erased any artifacts related to earlier coastal civilization by Native Americans. The site has seen many cycles of leveling, construction, and return to nature for nearly two centuries—not the least of which is the recent cleanup.

The remains of the old barge canal were found to have no historical merit. The study’s authors did regard as potentially significant the remaining buildings, structures and process equipment from the Arco Refinery period (1918-1936). Apparently, there are few remaining intact examples of commercial refineries from this early age. Since some ground pollution remains, and these old buildings do not contain vapor barriers, potential health effects from indoor air pollution must be considered in any future use scenario.

The Glynn Environmental Coalition received for review a Baseline Risk Assessment dated September 1997. The document contained a considerable amount of trivial material including arguments against the EPA' s risk assessment rules, discussions of EPA Region IV internal policies and many unusual variations of the normally expected models. After discussions with GEC, a letter was sent to EPA in which the report was wholly rejected as noncompliant. It required unreasonable expense on the part of the TAG (grant) program to comment on a line-by-line basis, since it was nearly impossible to extract fact from opinion. Apparently, EPA reviewers also found the document impractical and a major rewrite was obtained in October 1997.

This final RA version found that the site could pose a significant risk to any future residents. Further, the risks to site workers were "borderline" since the Hazard equaled 1 and the EPA regards as toxic anything with a hazard in excess of 1. Finally, seafood was also found to carry significant risk. All of the risk is non-carcinogenic risk.

The authors of the Risk Assessment (RA) conclude that there is no risk at the LCP site, and no further action is needed to ensure safety in the future. This conclusion is not a fair interpretation of the testing data.

Two different exposure models were used to test for both cancer and non-cancer risks (four different testing models); and, seven different use-scenarios were considered within the modeling framework (a total of 28 different results).

The two exposure models are called CTE (central tendency exposure) and RME (reasonable maximum exposure). Generally, RME is more likely to find risk than CTE, but both have advantages and disadvantages for interpreting results.

In the case of the LCP data, both the CTE and RME modeling systems predict no adverse risk for cancer. However, the models differ when predicting non-cancer effects. The CTE system predicts high Hazard values, but does not exceed EPA guidelines. The RME modeling predicts toxicity to future residents, primarily from PCB (a toxic chlorine compound) effects on human immune systems. Further, the RME system indicates that effects on workers are "borderline" in that the values map to 1, the EPA says that any value over 1 is significant.

EPA guidelines are not very specific for Hazard readings that calculate to precisely 1. Over 1, further Remedial Action is needed; less than 1 no further action is allowed. Right at the threshold requires examining other factors. The RA studies authors argue that PCB 1268 is not as toxic as other PCBs and the Hazard value of 1 is really an overestimation. They also argue that EPA’s models are too conservative in applying potential toxic effects. While there is some basis to their arguments it is also fair to say that the models can easily underestimate actual exposure, and the uncertainty in measuring toxicity is grounds for treating the chemicals as toxic, rather than safe.

Four is a very high value, indicating risk is likely. However, it is unlikely the area will be developed for homesites. The area is too industrialized to be practical as family dwellings in the foreseeable future. The EPA has expressed a desire to have the area placed under deed restrictions that prevent full time occupancy.

Although this value technically does not exceed the threshold value of 1 set by the EPA there are a variety of reasons it should be regarded as "borderline" with the likelihood that some risk is real. Converting a residential scenario into a worker scenario is mostly just changing occupancy from 24 hours a day, 7 days per week to 8 hours per day, 5 days per week. Some workers are likely to exceed the model and incur greater risk (their Hazard values would fall between 1 and 4). More importantly, some of the modeling systems averaged exposure over the entire site, but the PCB does not occur randomly. Some areas are more contaminated than others. Workers in these areas have greater potential exposure than at other portions of the site. There are some areas where workers will definitely have higher Hazard indexes of exposure than in other areas. These areas are clustered around the former facilities areas, which remain the areas most likely to be developed in the future.

Some areas still contain high levels of buried toxic waste. Any future development should include a health and safety plan that reduces all subsurface soil contact for future excavation workers.

The modeling for trespassers appears fair. Fences to restrict access from land border the site. Intrusion by boat occurs frequently but there does not appear to be significant contact with marsh soils.

The modeling used some unlikely scenarios—such as harvesting only from the area immediately adjacent Turtle River. This RME Hazard index is probably much higher and if valid data for the marsh were used, the CTE model would probably also predict risk from seafood. The Feasibility Plan should include funds for further enforcement and education regarding seafood harvest in this area. The harvest ban that has been in effect for several years should be extended indefinitely.

The LCP site is large and complex. The model greatly oversimplifies the site conditions, leading to bias in the results. In addition to the problem of site-wide averaging of toxins used in some of the RA modeling, the model averages "upland" or dry areas separately from "marsh" soils. Marsh access is presumed to be by boat only. It seems highly likely that future site workers will also have contact with the marsh at the shoreline. Any risks associated with the polluted shore do not appear tested in this Risk Assessment.

Also, the study’s authors discount any future use of remaining site buildings. However, the Cultural Resources review recommends retaining these structures. Studies on internal vapor concentrations are essential for any future use of the remaining buildings.

The Risk Assessment draft version found no Hazards exceeding 1, the threshold for further action. However, after EPA required additional modeling using more realistic values, both seafood and residential uses exceed the Hazard quotient of 1, and future worker scenarios mapped to the borderline values of 1. Since some of the calculations still contain questionable averaging practices, the Assessment and its conclusions should be regarded as biased.

The study’s authors provide an interesting, but not persuasive, debate for considering the PCB 1268 compound less toxic than predicted by the model. Also, the authors argue for relying solely on the CTE modeling—which is opinion, not science or policy. Experience at other sites indicates that values of 1 and over in the RME modeling system should be taken very seriously when considering potential long-term use of the site. The major remaining hazard appears to be PCB contamination. This contamination is concentrated in precise areas. The locations containing the contamination are also the areas most likely to be visited/used by on-site workers.

The Feasibility Plan is not yet available for review, so long-term planning is unknown at this time. Several groups have expressed interest in industrial development of portions of the site. However, the entire site is not free from risk as claimed by the authors of the Risk Assessment. Based on the RME modeling, development at this site could impact the safety of some workers and the community is justified in remaining concerned over site safety.