Such models are often a crucial part of the information analysis supporting regulatory decisions, an agency source says, because they help agency experts use existing data to "make decisions about future events." In some cases they have come under attack from industry groups that say models can be misleading or meaningless, leading to inaccurate projections of likely environmental outcomes and therefore unrealistic program targets and regulations.

The draft report could help EPA overcome such challenges by providing clear recommendations on how to use valid models and ensure they result in justifiable regulatory decisions, the source says. The report will help provide the agency with "guidelines and a vision of the selection and use of models" in future regulatory decisionmaking, according to the panel's Web site. The agency source says the draft NAS report will be sent to EPA at the end of this month.

EPA is also awaiting a report from NAS on how the agency should revamp its toxicological testing system for pesticides and industrial chemicals, which could replace the testing protocols EPA has used since the 1980s, a second agency source says.

NAS convened the scientific models panel in November 2003 at EPA's request to provide advice on ensuring models are of high quality and lead to justifiable decisions. A main impetus for that request was the 2001 passage of the Data Quality Act, the first EPA source says, which allowed outside parties to challenge the accuracy of information EPA disseminates for public use. Among the first challenges were a petition from the Competitive Enterprise Institute questioning the validity of EPA using climate change models as the basis for its 2002 global warming action report and a broad U.S. Chamber of Commerce challenge to EPA model validity.

"A crucial question when we started [requesting NAS' assistance] was how can we say our models lead to credible decisions?" the first EPA source says, adding that despite frequent uncertainties due to lack of data or untested hypotheses, "we have to make a decision" when regulation or program development is called for. The NAS report is meant to give EPA direction on how to ensure that use of models, such as contaminant dispersion models or epidemiological models, leads to justifiable regulations in the future, the source says.

The report will not be narrowly tailored to a single model or type of model, the source says. The panel members' expertise varies widely -- from law to chemical engineering to the history of scientific processes -- which will help the report give a range of suggestions and recommendations on how to think about information quality and how modeling results lead to decisions, as well as more technical aspects of the issue, the source says.

Once EPA has reviewed the draft, it will take the panel's recommendations and combine them with suggestions already offered by the agency's Science Advisory Board (SAB) last August to produce a new draft of the November 2003 "Draft Guidance on the Development, Evaluation, and Application of Regulatory Environmental Models," the source says. Among other recommendations, SAB said the guidance should more fully address "the need to develop and apply models within the context of a specific problem," should observe "[c]aution in the way that information on modeling uncertainty is evaluated and communicated," and should "more fully discuss uncertainty and sensitivity analysis methods." Relevant documents are available on InsideEPA.com.

Further, SAB noted in its review of the draft modeling guidance that the vision of the Regulatory Environmental Modeling (REM) program, which drafted the guidance, is "not matched by a commensurate, and steady, allocation of resources on the part of the Agency. It is therefore recommended that the Agency provide a meaningful commitment of resources to the REM initiative."

An EPA spokeswoman says in a statement that the agency "has taken the . . . guidelines very seriously and has put in the resources necessary to meet the spirit of the guidelines." The EPA Office of Research & Development "is funding the coordination expenses (i.e. seminars, workshops, and the centralized knowledge base)," the statement says without providing financial or other details.

The new guidance, when it is drafted, will be significant because "it articulates what we think of model evaluation vetting at EPA," the agency source says, predicting that it "will permeate throughout the agency." What is most important about the guidance is to have "an idea of what to do in evaluating models" that everyone can refer to, the source says.

The guidance will seek to describe how the principles of sound science should be addressed in vetting the use of models; to encourage peer review of a proposed model along with a data set to test its validity; and to examine how models meet the objectives and standards of quality assurance staff at EPA. Among the most important issues in the science of modeling is deciding which model could not only give plausible results but could best describe the real-world conditions researchers are attempting to understand, the source says. "You can come up with four plausible models," the source says, but it is important to decide "which one should you choose."

The source could not say when a new draft of the guidance would be available.
Meanwhile, NAS is also preparing to release a separate report on what a future system for conducting toxicological testing should look like. The report -- due to be released in the middle of June, according to an NAS source -- will seek to answer several questions EPA charged the committee with addressing in April 2004, including the benefits potentially arising from new testing technologies, the challenges to achieving them, and approaches and incentives that might be used to address the challenges.

The report will hopefully lay out a new testing system to replace the paradigm laid out in the 1980s for examining the toxicity of pesticides and the risks of industrial chemicals regulated under the Toxic Substances Control Act, the second agency source says. That system was created before many recent advances in data collection and analysis had been made, the source says, adding that the upcoming report will seek to lay out NAS' "vision" for an up-to-date way to take advantage of that progress.

Such questions are "very important to the pesticide program," for instance, because of the number of substances to be tested and regulated and because of the cost of doing expensive animal tests, the source says. "We shouldn't be asking for animal studies we don't need. The 1980s paradigm was good for that time, but we can design something better."

Among the newer tools likely to be discussed in the report are so-called quantitative structure activity relationship models, which can relate a chemical's biological activity, including its carcinogenicity, to its structural components, according to EPA's Web site. Such techniques are being increasingly used to identify chemicals and other substances most likely to be harmful to human health to narrow the field of potential subjects for more extensive tests, the source says.

An EPA spokeswoman did not respond to a request for comment on the report.