Comments on EPA’s Guidance for Performing Aggregate Exposure and Risk Assessments

Date: May 03, 2000

Conceptual Issues

EPA's new proposed guidance for performing exposure and risk assessments leapsgenerations ahead of the longstanding and well worn risk assessment methodologiesheretofore in operation in EPA's Office of Pesticide Programs (OPP). Amongst acacophony of complaints, something positive can be said for the Food Quality ProtectionAct of 1996 (FQPA) in that it has empowered EPA to make such improvements. The newGuidance is well written and well reasoned and avoids being too specific, therebyencouraging the assessor to use good judgement, but more effort should be invested inexplaining to EPA's rulemakers, managers and architects of new policy, who are notassessors, when, where and to what extent it should be applied.

To consider statistical surveys and probability distributions, mechanistic research,spatial and temporal linkages, bounding exercises, reconciliation of exposure and healtheffects, and sensitivity analysis, to mention just a few examples, and to guard againstscenarios that "defy logic," make the assessment process certainly morerational, but orders of magnitude more complex.

Being revolutionary rather than evolutionary is dangerous. Complex new concepts andprocedures cannot just be dropped into place. Chaos results. The people who must followthe new Guidance must have the training and experience to use it properly, or it will bemisinterpreted, and the infrastructure of the organization must be amended to beconsistent with it. Herein the Guidance document fails. It makes no effort to nest itselfproperly within the organization it is intended to guide, and unfortunately, signs ofchaos are already apparent. It is changing the organization that it intends to benefit,but without more effort that change will be permanently harmful.

The Anatomy of EPA's New Guidance

The guidance is written from the perspective of an assessor who is assigned the task ofanalyzing the aggregate exposures and risks associated with the use of a pesticide that isalready widely registered. It is NOT written as guidance for rulemaking or for determiningregistration data requirements.

For existing chemicals, the assessment problem is approached like a stepwise researchproject where scientific method is applied. First, all existing data and availableknowledge are reviewed and organized. Mini-theories are then generated to guide theresearch. If one uses "assessor" language, each mini-theory is called an"exposure scenario". Each theory is like a null hypothesis that says, in effect,"Exposure scenario #231 (or #b171, or #144, etc.) produces significant risk." Ifthe null hypothesis is shown to be untrue, it is rejected, i.e., analyzed nofurther.

In practice risk assessment is, of course, more complex. The research project uses aseries of experiments to test each theory (i.e., each exposure scenario) in ascientifically rigorous and logical stepwise manner. In the early steps of theinvestigation, use is made of available data to construct rough bounding estimates ofexposure and risk.1  The early analysesuse exceedingly conservative assumptions so that the calculations are unlikely tounderestimate risk (i.e., highly likely to overestimate risk drastically). Boundingestimates can be quite simple when it is obvious either: 1) that the risks aresubstantial, or 2) that the risks are negligible.

If during the bounding exercise, or for that matter at later points, negligibleexposure and risk are found for a particular theory (i.e., exposure scenario), itsassociated exposure pathway is examined no further.2  Unlike a null hypothesis, however, the theory is not entirely discarded. Carefulrecords are kept of the analysis and conclusions. Even though exposure scenarios in theearly stages of the analysis may be so simple minded that they obviously do not representreality, they may work to provide useful predictions that certain types of exposureproduce negligible risk. Like economic theories that seem obviously too simplistic, e.g.,the theory of supply and demand, such scenarios are exceedingly useful components in theoverall assessment.

If the bounding estimates suggest significant exposures and risks for a particularexposure pathway, the subsequent steps in the analysis of that pathway become moredetailed. The Guidance document necessarily spends most of its effort describing thesedetailed methods of analysis and how to combine the analyses for a collection of exposurepathways into an aggregate. An integral part of that discussion involves the types of datathat are needed as the assessment becomes more complex.

When developing an assessment for a pesticide that is already widely registered, forexample during reregistration or tolerance reassessment, a substantial body of data andpractical use experience is generally already available, and EPA has authority to ask theregistrant to provide additional key studies. Efficiency is built into the process in thatEPA approaches the analysis from a rich knowledge base and does not ask for new data thatare obviously unnecessary. Likewise, the Guidance instructs that the assessment should goonly to the level of detail necessary to substantiate that the pesticide can be usedsafely or that risks associated with a certain exposure pathway need to be mitigated. TheGuidance is careful to point out that aggregate exposure and risk assessment forregulatory purposes is not an academic exercise to be pursued like basic researchto its ultimate level of intellectual accomplishment and complexity.

Troubling Signs of an Organizational Problem

The bottom line -- In embracing the new Guidance without proper preparation, EPA's assessment practices and data requirements for registration are quickly becoming wildly incompatible with the conceptual framework upon which the Guidance rests.

-- In embracing the new Guidance without proper preparation, EPA's assessment practices and data requirements for registration are quickly becoming wildly incompatible with the conceptual framework upon which the Guidance rests.

EPA's new Guidance is based on experience with a conceptual framework that haslong been followed by those knowledgeable in the field of exposure and risk assessment.EPA's formal public policies dealing with pesticides before FQPA were much lesssophisticated, employing in most cases point estimates of risk calculated using simplisticstandardized procedures with numerous built in worst-case assumptions. The new Guidancedocument, in fact, closely describes assessment practices unofficially followed for yearsby OPP's most senior scientists when particularly difficult special reviews andreregistrations were encountered. The new Guidance formally embraces the use of new typesof data, but these types of data were considered before when available and needed.

If the Guidance is not really new, why is there a problem? We believe that an unhealthyholiness has been imparted to the full set of recent exposure and risk assessment guidancedocuments by the fact that they were prepared as part of the Tolerance ReassessmentAdvisory Committee (TRAC) process, a process created by EPA's Administratorostensibly to provide the ultimate "keys" for implementing FQPA. As aconsequence, the Guidance is being used in ways that go beyond its scope.

Taken out of context, this aggregate assessment guidance document, like others, seemswell done. But, when viewed from the perspective of its implementation within EPA'sorganization, often by staff with scant or no experience in complex risk assessments, datarequirements are proposed or applied in ways that make little sense and complexassessments are constructed where they are clearly unnecessary. Policy makers in allcorners of OPP have for some time been slavishly revising data requirements and operatingprocedures to comport with the TRAC documents, leaving an abundant record that shows alack of understanding of the science behind such assessments.

Unlike an assessment of an already registered chemical, the use of the Guidance toupdate data requirements and procedures has horrendous built in inefficiencies. For thattask the Guidance must go further, because policy makers seem unable to conceptualize, ina generic sense, when a particular study should or should not be required and when anexposure scenario does or does not need to be assessed at length. If FQPA implementationcontinues to march along without such critical underpinnings, the Guidance will lead usinto a future governed by rules that require all possible data all of the time, andpolicies that direct EPA scientists to assess mindlessly every imaginable, and some not soimaginable, exposure scenarios to the "nth" degree.

Mis-implementation is already having serious consequences. Case in point, EPA hasinformally released an updated set of data requirements for antimicrobial pesticides, 40CFR 158 Subpart W, which will soon be proposed formally. We have heard that Subpart W isthe first in a series of planned updates in the data requirements that are being designedto implement FQPA. The new Subpart W requirements correspond almost exactly to the typesof data described for very complex exposure and risk assessments in the recent guidancedocuments produced by the TRAC process. Unlike existing codified data requirements, mostof the requirements in Subpart W are not conditional. Furthermore, arguments regardingtoxicity or exposure (or common sense) are no longer considered relevant when seekingwaivers of data requirements for products that fall under Subpart W. The only ground forwaiver considered relevant is that it is physically impossible to conduct a study.

Another case in point, EPA is finding it impossible to approve new inert ingredientsbecause it cannot complete aggregate exposure and risk assessments for non-pesticidal usesof such substances, even when there are clearly no questions of safety. The fact that asubstance has been used widely and safely for years in a variety of non-pesticidalproducts is viewed as a huge problem, not an assuring indication of safety. The regulatoryprocess is stalled because a full "FQPA assessment" cannot be constructed.Exposure scenarios dealing with the non-pesticidal uses of such substances invariablycannot be enumerated in any detail. New policies hold that worst-case bounding assessmentsshowing negligible exposure or risk are not adequate to satisfy FQPA assessmentrequirements. Rather than enlarging such policies to include the use of bounding estimatesand common sense, EPA is now working diligently on a generic expansion in the toxicologyand exposure data requirements for inert ingredients. These expanded requirements willsoon be applied to the tolerance reassessment of inert ingredients. It will be the raresubstance, indeed, that will arrant the investment needed to satisfy the data demands of afull assessment.

One final case in point, with particularly sad consequences, is the likely impact onEPA's much lauded reduced risk pesticide initiative. A key element in OPP's"mission statement" is the promotion of the safer use of pesticides. Towardsthis end EPA has encouraged development of new safer pesticides by offering expeditedreview. It is in the public interest that such pesticides should be given preferentialtreatment. In concept, the review process can be shortened because less data are requiredto assess the risks of substances that are obviously very safe (e.g., show no toxiceffects at limit doses). Time is also saved because exposure and risk assessments can besimplified for substances that have little or no toxicity by any feasible route ofexposure.

Tragically, the thinking that made the reduced risk pesticide initiative possible, andso appealing to Congress, is incompatible with current implementation of the Guidance.Applicants seeking to register reduced risk pesticides are now being asked to prepareformal aggregate exposure and risk assessments to comply with "the requirements ofFQPA," even when toxicity and exposure are obviously negligible. EPA'simplementation of policies that standardize data requirements and risk assessmentprocedures portends the disappearance of the reduced risk initiative. Reduced risk is acomparative property. Some pesticides are much safer than others. Unfortunately, inherentin EPA's new policies is the concept that it is not possible to discern suchdifferences among pesticides.

Returning to the bottom line, the future will fast be upon us where EPA'spesticide program no longer follows anything that resembles a scientific process. NewGuidance has been inserted into the organization like a defective seed from whichdistorted new policies and regulations are rapidly growing.

EPA must with a sense of urgency take action to make the data requirements, policiesand procedures within OPP consistent with the Guidance. The data required for the mostelaborate of assessments is not required by the Guidance for every situation. EPA staffare demanding unnecessary data to support an ever increasing number of resource intensiveassessments for trivial exposure pathways and non-toxic chemicals. As a consequence,EPA's pesticide regulatory program is in danger of losing all sense of priority.

1 See the excellent discussion of the early part ofthe analysis on pages 29 through 32 of the Guidance document.




2 Before concluding that risks are negligible, a sensitivityanalysis may be performed as described on pages 39-40 of the Guidance document. Thedescription reads in part: "Sensitivity analyses are particularly useful in decidingwhether or not to evaluate a pathway-specific analysis to the next level of datarefinement and therefore consume more resources."