To E-mail Recipients: NOTE: The following are ASCII E-mail versions of an MS WORD letter and its attachments and contain the same information though not in as a nice a format. The ASCII E-mail format has been chosen as an additional distribution method because of efficiency and ease of transport to all receivers. If a recipient can process an attached WORD file and desires the MSWORD version via E-mail or the hard copy version, please feel free to request it. There are Internet sources (URLs) on another page. A. A. Brooks ================= Letter =============== Date: March 12, 1997 Dear Senator or Representative: Subject: Senate Bill S.8 - The Superfund Cleanup Acceleration Act of 1997 The current CERCLA laws repeatedly require cost effective remediation of Superfund facilities and should have served to bring about unbiased, accurate risk assessment. Yet these laws have resulted in the current procedures for risk assessment which require the use of conservative data and models. These in turn lead to the costly remediation of facilities that are clearly not a significant risk to public health nor to the environment. While this is stated to be "best science", clearly it is not, as the data used intentionally lead to risk estimates that are conservative by a factor of 1,000 or more. Risk assessment that is this wide of the mark cannot be considered as best science nor can it lead to cost-effective remediation. This biased implementation of the CERCLA law has been justified as "a mandate from Congress to restore CERCLA facilities to an unquestionably safe level" and that "it cannot be changed unless the law is changed." If such a biased and long-standing implementation also is not to be read again into the current CERCLA reauthorization, then the new wording must be carefully crafted to preclude it. The present wording of Senate bill S8 is good but not impervious to distortion. It should be further strengthened to preclude the same biased interpretation of the risk assessment process as the previous CERCLA laws have permitted. I am suggesting a small but important strengthening of the wording in Senate bill S.8 to preclude the continued implementation distortions that have occurred in the past and which have resulted in so many justifiable criticisms of CERCLA. Based on a knowledge of the science and from my intimate experience with CERCLA processes over the last seven years, I sincerely believe that these wording changes will be necessary to bring realism and cost-effectiveness into the remediation process. The current EPA administrative reforms to expedite the process, while desirable, will not effect the necessary fundamental changes to the risk assessment model. The attached suggested wording changes to S.8 are intended to do just that: to preclude the use of biased data in the risk assessment process and to present to the public and risk managers the best unbiased estimates that the science of risk assessment has to offer. Also attached (for the use of the technically inclined) is a brief lay description of the assumptions of the current practices and the more realistic assumptions of the practices I believe Congress intends. The misconceptions of the "always safe error" assumption with limited funding are described. If it is required, I can supply more detailed background information and technical description. I would appreciate hearing if you can support all or any part of the suggested wording or would support other wording for the same purpose. Very respectfully yours, ___________________ Alfred A. Brooks 100 Wiltshire Drive Oak Ridge, TN 37830 Ph: (423) 482 1559 Attachments (2) ============= Attachment 1 ============================ Suggested Wording for Senate Bill S.8 Superfund Cleanup Acceleration Act To Ensure Compliance with Congress' Intent These suggested changes to the wording of Senate Bill S.8 - Superfund Cleanup Acceleration Act of 1997 are intended strengthen the current language so as to ensure that the intent of Congress for the "cost- effective remediation" of "substantive health risk" are indeed reflected in the regulations, procedures and methodologies which result from this Act. The wording is intended to replace the current "always safe error" assumption which is equivalent to "unlimited funding" by the more reasonable assumption of "limited funding" which calls for unbiased, realistic data and risk analysis methods suitable for the optimal allocation of funds to the many Superfund facilities. Base Document: S.8 - Dated: January, 21, 1997 (Stapled version with 259 total pages and line numbers) Section 301, page 68, after line 13: Insert an additional paragraph [1] as follows: "(viii) Members of the general public who have technical knowledge of the facility or the technical areas of the risk assessment or remediation methods which may be required at the facility. Section 402, page 85 line 7: After [2]: "cancer" insert "risk". Section 402, page 85 line 9: Replace [3] "range from 10-4 to 10-6" by "less than 10-4" Section 402, page 85 line 15: At the end of (II) after "hazard index of 1" insert [4] "where the hazard index is based upon a human reference dose which is realistic, unbiased and statistically significant." Section 402, page 103, line 16: After "objective," insert [5] "statistically significant," Section 403, page 104, after line 8 at the end of "b) RISK EVALUATION PRINCIPLES" Add additional paragraphs [6] as follows: "(5) be based on accurate, realistic risk assessment which is free of non-technical objectives and biases and whose risk levels are commensurate with other significant societal risks; (6) shall reflect the need to allocate the available funds to the several sites so as to reduce the total public risk to a minimum for the limited funds available; (7) shall properly accommodate the statistical variation in the target populations and the data, and shall protect a reasonably high percentile of the exposed population; (8) shall be based on an information base of unbiased data, and guidelines for unbiased models to effect the foregoing objectives; (9) shall present several realistic, alternative remediation goals and their estimated costs to the risk managers and public for final selection." Section 403, page 105, line 21, at the end of "(c) RISK COMMUNICATION PRINCIPLES" after list item (5) Insert an additional paragraph [7] as follows: "(6) issue health consultations which shall state the maximum contaminant levels and conditions for which the site can be considered safe based on an unbiased risk assessment." Section 904, page 255, line 23: After "technologies" insert [8] "including unbiased, realistic risk assessment". Footnotes: [1] The purpose of this wording is to ensure that the public has the benefit of non-affiliated local experts to interpret and aid in technical aspects of the ongoing process and that proper site-specific risk assessment models are used. [2] Clarity [3] The background lifetime risk for all cancer is about 0.3. Protection to a level 300,000 times lower does not seem cost effective unless the cost of such remediation is exceedingly low. [4] To ensure that conservative biases are not introduced through the human reference dose. [5] To prohibit the use of epidemiological results which do not meet the accepted scientific criteria for validity. [6] To ensure that the risk assessment is carried out at the state of the art with the best unbiased data available as well as producing results which are commensurate with other societal risks and which are suitable for the near-optimal allocation of resources to obtain cost- effectiveness. These requirements should define and ensure the use of the best risk assessment science. [7] To insure that the expertise of ATSDR can be brought to bear on unbiased risk assessment in an timely manner thus reducing delays and costs. [8] To provide for the preparation of an unbiased risk assessment data base. ======= Attachment 2 ========== CERCLA Re-Authorization As the re-authorization of CERCLA by Congress approaches, there is a unique opportunity to review the basic assumption EPA has used in risk assessment and to effect the changes deemed necessary to attain the realism in risk assessment needed to properly allocate our limited resources to Superfund sites and to make the difficult choices between remediation alternatives. The problem, in its simplest terms, can be presented as a choice between the two cost models described below: The Current Unlimited Resource Model EPA has described CERCLA as requiring that they propose remediation goals that will leave a site "unquestionably safe" and this in turn justifies the use of very conservatively biased data and exposure models. As long as unlimited resources are available to Superfund and hence to each remediation site then the risk assessment model may be conservatively biased and the input data may consist of the conservative bounds on the data. The assumption of the "always safe error" is valid, the model admits the error of repeated conservative statistical approximations and eventually the sites will be cleaned up to an "unquestionably safe" level as EPA desires. The EPA appears to view the other government agencies as bound by their rules and models. Excessively conservative bounds will always lead to unnecessary remediation costs unless the lay public is prepared to contest the expenditures on a site by site basis at the risk management phase. Seldom can the lay public refute the conclusions drawn from the conservative, complex risk assessment when faced with professionals who are required to defend an assessment they know to be inaccurate and biased. The Proposed Limited Resource Model On the other hand with limited resources, optimization theory, which deals with the efficient allocation of limited resources to multiple tasks, requires that there be an optimal allocation for which the maximum remediation is obtained and that all other allocations are less effective and thus riskier. In the best of all worlds, this model would consider all the sites together, demand a lack of bias in the risk assessment data such that it reflects reality insofar as possible. The approximations of the real world, particularly the need to remediate sites separately, do not invalidate the existence of the optimal allocation nor the lessons to be learned form the model. In this model, there is no such thing as an "always safe error" and any statistics must be properly applied to avoid errors and less than optimal solutions. The resources will be allocated as best we can to each site in order to reduce the final collective risk as much as possible. The lack of suitable data does not negate the conclusions drawn here but reduces the ability to accurately approximate the optimal solution; deliberate large, so-called conservative biases in the data lead to increasingly poor solutions. Simplistically, to avoid error one must try to be accurate. This model, giving rise to realistic alternatives, allows the public to make well-founded input to the process and select between alternative, proposed remediations. Concerning the major premises: Limited Resources vs. Unlimited Resources. While unlimited funding may have appeared plausible at one time, it no longer seems tenable in the face of the existing multitude of Federal fiscal problems. Limited resources seem to be much more realistic for the future. CERCLA now says "significant health risk" and "cost effective remediation". This sounds more like limited resources than unlimited resources, yet the EPA model is suitable only to unlimited resources. What wording in the new legislation must Congress employ to change the models used by EPA to reflect the limited funding realities of the present and the intent of Congress? Wording such as the following is suggested: "The remediation proposed by the CERCLA process shall be based on accurate, realistic risk assessment which is free of political objectives and is commensurate with other significant societal risks, and shall reflect the need to allocate the available funds to the several sites so as to reduce the total public risk to a minimum for the limited funds available. The CERCLA procedures shall accommodate the statistical variation in the target populations and the data, and shall protect a reasonably high percentile of the exposed population. The EPA shall provide an information base of unbiased data, and guidelines for unbiased models to effect the foregoing objectives. The risk assessment phase shall present several realistic, alternative remediation goals to the risk managers and public for final adoption." Optimal Allocation - A Simple Example Optimization theory deals quantitatively with the efficient allocation of limited resources to multiple tasks with the intent to obtain the maximum return for the resources available, i.e., the "most bang for the buck". Optimization theory is not new; it dates back to about 1800, with additional progress occurring in the 1950's and 60's by using computers to solve a wide range of problems. The details of optimization theory involve sophisticated mathematics, but the following analogy illustrates the application of optimization theory in lay terms to a problem easily visualized. Assume you wish to climb to the top of the highest peak in a complex mountain range for which there are good maps, good aerial photographs and clear weather allowing you the study the mountains by telescope. To your dismay you determine that you can not possibly get to the top in the time available before your vacation ends, and your boss won't give you more time off. You decide to climb as high as you can in the time allowed, and by studying the maps, photos and the terrain, you determine which of the many possible routes will take you as high as possible in the time you have. In optimization theory, this route, #1, would be called the optimal strategy, i.e., the optimal allocation of the available time. All other possible routes would be terminated at a lower altitude when the time ran out. See diagram. /^\ Assent routs (3) ****** #n / \ #1 * / \\ Route #1 reaches the highest altitude * / \* #2 and is the optimal strategy. * / *#3 \* * / * \* * / * \* Three Ascent Strategies Now consider there is an identical mountain range except that it is always shrouded in clouds so there are no accurate maps nor photos nor telescopic views. Another equally skilled climber (in a moment of weakness and not knowing it is identical to the above range) has selected this mountain range to climb under the same time constraints. Since the mountains are identical, the same "best" route exists (even in the absence of good information) and should be used. But our second climber has little chance of reaching the same altitude as the first unless he is very, very lucky. The lack of information did not change the problem but only forced the climber to make poorer choices on how to solve it. Note: Either mountain could have been climbed to the top if the boss had allowed enough time off; probably a lot more for the second climber. Note also that it is the total route that is the "best" and that time wasted at any point in the route will reduce the total altitude gained. To complete the analogy, the several parts of the "best" route are equivalent to the several Superfund sites among which the resources must be allocated. The lack of terrain information is equivalent to the badly biased risk assessment data used to set allocations to Superfund sites. The boss of course is Congress who can not or will not allocate unlimited funds in the face of other demands. With limited resources an optimal strategy or allocation of resources always exists. Ignorance of the problem does not negate the conclusions that could be drawn from optimization theory, and the lack of good information will lead to less than optimal allocations and wasted resources. In the case of Superfund sites, this means more residual risk when the funding runs out. ===============================