Tobacco Documents Online

the agency makes a judicially reviewable decision. It is a post-hoc rationalization that serves only the Agency's litigation interests. 2. Discovery Is Necessary Because EPA's "Record" Is Incomplete and Does Not Provide Adequate Explanation of the Agency's Decision The decision in this case should be made on "the full administrative record that was before the [Agency] at the time [it] made [its] decision." Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 420 (1971); see also Appalachian Power Co. v. EPA, 477 F.2d 495, 507 (4th Cir. 1973) ("in making its review, the Court must have . . . 'the whole record' on which the agency acted"). The "whole record" includes "all documents and materials directly or indirectly considered by the agency decision-makers and includes evidence contrary to the agency's position." Thompson v. United States Dep't of Labor, 885 F.2d at 555. Where, as here, the completeness of the record is an issue, the Fourth Circuit has allowed discovery after plaintiffs' preliminary showing that: [T]he administrative record as presented may be incomplete and . . . there may be evidence or documents which were available for review by the agency but not reviewed, or which were reviewed and not relied upon by the agency, or which were reviewed and relied upon by the agency but not contained in the "administrative record." Franklin Savings Ass'n. v. Ryan, 922 F.2d 209, 210 (4th Cir. 1991). In such circumstances, the plaintiff was permitted to - 11 -

obtain the production of documents and to take depositions of agency employees.8/ See also LTV Steel, 119 F.R.D. at 343 (court allowed deposition of agency's executive director as to "what matters, whether discussed orally or committed to writing, were considered in the course of the [agency's] deliberations"); Petrolane, Inc. v. Dept. of Eneray, 79 F.R.D. 115, 119 (C.D. Cal. 1978) (court allowed plaintiff to "discover any materials needed to complete the administrative record").9/ We already know from the Farland Declaration that the "record" which EPA would have the Court review does not contain all the information before EPA when it made its ETS decision. What follows are additional examples of what information and explanations we also know to be missing from EPA's inadequate "record." Only discovery will disclose whether there is other missing information. 8/ Discovery is the appropriate remedy here "to provide a record of all documents and materials directly or indirectly considered by the agency decisionmakers." Public Power Council v. Johnson, 674 F.2d 791, 794 (9th Cir. 1982) (Kennedy, J.) (citing, inter alia, Appalachian Power, 477 F.2d at 507); see also Dopico v. Goldschmidt, 687 F.2d 644, 654 (2d Cir. 1982) ("defendants' assurances that they have submitted the full record will not substitute for the Court's independent consideration of that issue after some opportunity for discovery"); Natural Resources Defense Council, Inc. v. Train, 519 F.2d 287, 292 (D.C. Cir. 1975) (parties challenging agency action are entitled to discovery to learn whether "documents which are properly part of the administrative record have been withheld"). 9/ To obtain discovery, plaintiffs need merely "demonstrate some reasonable basis for believing that the administrative record is incomplete." Texas Steel Co. v. Donovan, 93 F.R.D. 619, 621 (N.D. Tex. 1982); see also Petrolane, 79 F.R.D. at 119 n.12 (discovery may be obtained where "a showing is made that [the record] may not be complete").

a. ECAO-CIN Memoranda and Discussion of Group A Versus Group B. In an apparent effort to omit some particularly damaging documents from its "record," the Farland Declaration creates a self-serving distinction between materials EPA "relied upon" (1 11), which are in the "record," and the deliberative materials EPA merely "considered" (1j 14), which are not. By this maneuver, EPA banished from the "record" a series of internal memoranda that blow the whistle on EPA's Group A classification. The excluded memoranda were created by the Agency's own Environmental Criteria and Assessment Office in Cincinnati (ECAO-CIN). This office is responsible for reviewing, among other things, EPA's risk assessment documents on air pollution issues. See Declaration of Larry R. Glass, Ph.D. (Glass Declaration) 1 3, attached at Appendix Tab 3. When the ECAO-CIN office reviewed the 1990 and 1992 drafts of the ETS Risk Assessment, it concluded both times that EPA's Group A classification was not warranted. Though EPA excluded the 1990 and 1992 ECAO-CIN memoranda from its "record," we have provided them for the Court's consideration as Exhibits B and C to the Glass Declaration (Appendix Tab 3). EPA's "record" not only excludes these documents without even noting their existence, it also fails to provide responses to the criticisms raised and the suggestions made by these reviewers. EPA simply ignored them and would have the Court do so, also.

In 1990, Dr. Glass was the ECAO-CIN toxicologist charged with reviewing the draft Risk Assessment. He and other EPA professional reviewers expressed "concern with the classification of ETS as a Group A carcinogen." Glass Declaration 1 7. In fact, Dr. Glass specifically recommended against the Group A classification in 1990. Id. 1 6. In 1992, the ECAO-CIN office again challenged the Group A classification as failing to satisfy the Agency's Guidelines for Carcinogen Risk Assessment.lo1 ECAO-CIN told EPA headquarters to substantially revise the risk assessment and to incorporate: ... a full discussion of carcinogen category A vs. B, based on the absence of definitive data of [ETS] in humans. Like it or not, EPA should live within its own categorization framework or clearly explain why [it] chose not to do so. Glass Declaration ~[ 9. Despite these repeated admonitions, the final ETS Risk Assessment was not substantially revised. The document does not address or respond to ECAO-CIN's devastating criticisms. See Glass Declaration ~[ 10 ("There is no analysis in the final ETS document as to whether Group A versus Group B, for example, is the proper classification category for ETS ..."). We know ECAO-CIN's concerns registered in Washington, because Dr. Glass received an angry telephone call from Dr. 10/ The Guidelines stipulate that the Group A (known human) classification can be "used only when there is sufficient evidence from [human] epidemiologic studies to support a causal association between exposure to the agents and cancer." 51 Fed. Reg. 33992, 34000 (Sept. 24, 1986). - 14 -

Bayard, the ETS Project Manager. Id. 1 8. But there was never any substantive response. Id. Discovery will shed light on why and how Washington chose to ignore ECAO-CIN's comments. EPA's "record" also does not explain why the Agency failed to address these same concerns when they were raised by members of the Agency's Science Advisory Board (SAB). In 1990, Dr. Geoffrey Kabat, an SAB member, said that "classifying ETS as a Class A carcinogen is maybe a little rash." "Admin. Rec." I.SAB.16.2 at 15 (transcript of SAB public meeting, December 5, 1990). In 1992, he again said that he was "uncomfortable" with the Group A classification. "Admin. Rec." II.SAB.9.16 at 1 (comments of Geoffrey Kabat). Dr. Kabat also questioned EPA's completeness and its even-handedness. He commented as follows on the EPA 1992 draft: I am still uncomfortable with certain aspects of the treatment . . . of passive smoking and lung cancer, which, in my view could still be presented in a more even-handed manner. A number of inconsistencies in the epidemiologic studies are ignored, and other issues relevant to the assessment of the carcinogenicity of ETS go unmentioned. Id. Dr. Kabat was not alone on the SAB in questioning the propriety of the Group A classification. Dr. Victor G. Laties stated his opinion too that EPA had not yet demonstrated a justification for the Group A classification. "Admin. Rec." I.SAB.23.14 at 6 (comments of Victor G. Laties). These criticisms of EPA's classification decision paralleled Z\Z ~ those of ECAO-CIN, yet they were apparently also ignored by EPA. ~ F.~ - 15 - r> ~

EPA's response to these concerns, if any, must be examined in order to understand and evaluate the Agency's decision to classify ETS as a Group A carcinogen. Discovery will determine why EPA ignored the serious issues raised by these reviewers, including the all-important question of Group A versus Group B (or other Group) under the Risk Assessment Guidelines.ll/ b. 95% Versus 90% Confidence Interval. The 1990 draft of the ETS risk assessment employs the 95% "confidence interval" to evaluate the epidemiologic studies regarding a possible relationship between ETS and lung cancer.12/ This is the standard statistical model that scientists, EPA, and other agencies traditionally use to avoid the possibility that a perceived causal relationship is really due to chance.13/ In the 1992 draft, however, the confidence interval was relaxed to an unconventional 90% level. This statistical shift made it easier for EPA to claim a link between ETS and lung cancer. 11/ As the Court has noted, one of the major issues in this case is "whether EPA followed its internal guidelines in formulating the Report." Mem. Op. at 17. 12/ "Admin. Rec." ORD.B.2 at 1-4, 3-18 n.l, 3-19 (1990 Draft ETS Risk Assessment). 13/ See Rothman, K. J., Modern Epidemiolocty, "The Role of Statistics in Epidemiologic Analysis: Assessment of Random Variability," Little, Brown & Company, Boston, 1986, at 116-117; see also N. E. Breslow and N. E. Day, Statistical Methods in Cancer Research, International Agency for Research on Cancer (Lyon, France, 1980), at 128 ("conventional significance levels" are 99% or 95%). The ETS Risk Assessment references Breslow and Day as one of its principal statistical authorities. See, e.g., Final Risk Assessment at 5-14. - 16 -

The change in confidence intervals was not highlighted in any document or explained to any reviewers, including the SAB. Nonetheless, the SAB's Dr. Kabat saw the problem and admonished EPA that the use of a 90% confidence interval (instead of 95%) "is to be discouraged." "Admin. Rec." II.SAB.9.16 at 6 (comments of Geoffrey Kabat). Dr. Kabat characterized EPA's statistical sleight-of-hand as follows: "It looks like [an] attempt to achieve statistical significance for a result which otherwise would not achieve significance." Id. From the "record" EPA offers the Court, it appears that EPA did not address or respond to Dr. Kabat's concern. EPA went with the weak 90% confidence interval in the final ETS document without ever explaining the impact of relaxing the standard. In fact, EPA had to manipulate the underlying epidemiologic studies in order to compute their results at a weaker 90% level. Of 30 studies that EPA relied on, 27 reported their own results at the standard 95% level. Complaint 5 31; see also Final ETS Risk Assessment, Appendix A. The only place EPA reports its 95% to 90% "conversion" is in the second sentence of a footnote, n.3, to a table buried on page 5-27 of the final ETS Risk Assessment. It reads: "All corrected 95% confidence intervals have been converted to 90% confidence intervals." That is the extent of EPA's discussion of its statistical manipulation -- it fails to justify the change. While the "record" does not discuss this statistical legerdemain, EPA disclosed its motivation at a congressional - 17 -

hearing on July 21, 1993.14/ In response to a question from Congressman Rose, Dr. Bayard, EPA's ETS project manager, disclosed that when EPA applied the standard 95% confidence interval, it came up with a relative risk that included 0.99: this meant there was no statistically significant correlation between ETS and lung cancer.15/ Id. at 40. In order for Dr. Bayard to confess this fact to Congress, EPA obviously performed calculations with the 95% confidence interval that were not reflected in the Risk Assessment. But where are the relevant worksheets? We do not find them in EPA's "record." Because the Risk Assessment Guidelines require the Group A classification to be used "only when there is sufficient epidemiologic evidence," one can imagine why EPA is trying to sweep its 95% confidence interval information under the rug. Discovery will identify precisely how EPA's decision was influenced by its statistical manipulations. c. Relevant Studies Were Excluded. Another significant gap in the "record" EPA crafted for this litigation concerns EPA's choice of which epidemiologic studies 14/ Review of the U.S. Environmental Protection Agency's Tobacco and Smoke Study: Hearing Before the Subcomm. on Specialty Crops and Natural Resources of the Comm. on Agriculture, 103d Cong., 2d Sess.; excerpt attached at Appendix Tab 4. 15/ A relative risk of 1.0 or less means that there is no positive association between the risk factor and the disease. See, e.g., Breslow & Day, Statistical Methods in Cancer Research, supra, at 128. A confidence interval that includes 1.0 within the upper and lower bounds of its range is considered to be not statistically significant.

to include in its evaluation of ETS. The final Risk Assessment states that two studies, Brownson et al. and Stockwell et al., were excluded from EPA's consideration because they were published after EPA's arbitrary cut-off date, but well before the final Risk Assessment was completed. See Final ETS Risk Assessment, Addendum at ADD-i. Inclusion of these studies in EPA's analysis would have further weakened the alleged statistical correlation claimed between ETS and lung cancer. Complaint f 56. While EPA failed to consider the Brownson and Stockwell epidemiologic evidence, it went out of its way to include an incomplete study, by Fontham et al., whose results appeared to be more favorable to EPA's desired conclusion. Not only was the Fontham study incomplete when EPA included it, but the "record" reveals that EPA's consultant actually importuned the study's author to release data to the Agency. See "Admin. Rec." ORD.E.2.17 (Kenneth G. Brown letter to Elizabeth T. H. Fontham, October 12, 1991).16/ Why was it so important to press Dr. Fontham for her incomplete data, but not so important to postpone 16/ Mr. Brown's October letter to Dr. Fontham stated: ... your interim study report has created a lot of interest at EPA. They would like for me to include it in our report ... We are aware of your November publication date Id. - 19 -

the arbitrary cut-off date so that the results of Brownson and Stockwell could also be incorporated? The ECAO-CIN office also asked about missing studies in its 1992 critique: What is lacking is a clear statement of why these particular studies were selected for review. Was any attempt made to include non-published studies (which are likely to have non-positive findings) in the review? ...[I]t is not really clear to me how the process of data abstraction took place. * * * It seems that several studies were excluded a priori from the meta-analysis for evidently different reasons. It is mentioned that the study of Sandler et al. was excluded [due] to a small number of lung cancer cases -- I thought the purpose of pooling data was to include studies such as these. If it was totally useless, this should be more clearly stated. 17/ The "record" also contains one mysterious reference to a draft ETS risk assessment that was prepared by EPA in 1988. See "Admin. Rec." ORD.E.3.24 (letter to Hon. Thomas J. Bliley, Jr., from Gary J. Foley, Acting Assistant Administrator for Research and Development, undated, attaching copy of contractor "work assignment" form, February 14, 1989, accompanying memorandum at 7). This document, too, is neither included nor explained in the "record." 17/ ECAO-CIN Memorandum from Patricia A. Murphy to Lynn Papa, "Requested Review of OHEA documents on Passive Smoking Health Risk Assessment," March 23, 1992, at 4 (exhibit C to Glass Declaration at Appendix Tab 3). - 20 -