Submit to Digest

Burdens of Discovery for Scientific Working Materials and Deliberative Documents

Commentary Notes First Amendment
Written by: Evelyn Y. Chang Edited by: Jessica Vosgerchian [caption id="attachment_3299" align="alignleft" width="83"] Photo By: Horia Varlan - CC BY 2.0[/caption] In March of 2012, British Petroleum sought court enforcement of a subpoena for “any conversation or discussion” made by researchers from the Woods Hole Oceanographic Institution (“WHOI”) regarding their studies on the Deepwater Horizon oil spill. WHOI and its researchers were neither parties nor witnesses in the ongoing lawsuits related to the Deepwater spill, but they had contributed data to a government report on the spill, and results of their studies had been cited in the lawsuits. WHOI argued that they had already provided all the information needed to test the study’s veracity, but BP asserted that additional materials, including records of internal deliberations made by the researchers before they published the studies, were necessary for them to challenge the studies’ results. The court applied a balancing test that weighed BP’s need for the requested information against the burden placed on WHOI, and required the WHOI researchers disclose internal pre-publication materials relating to the studies cited in the government report.[i] Last fall, Science published an editorial by several WHOI researchers whose materials were among those requested by the subpoena. The WHOI researchers argued that BP’s subpoena request amounted to an attack on the independence of scientific inquiry, and advocated for federal legislation to protect researchers from legal harassment when their results contradict entrenched interests. While legislation is the most straightforward method by which the government can protect scientific independence, common law already provides some protection for researchers who may find themselves in a situation similar to that of WHOI. Discovery of evidence from third parties is nothing new, and many courts hold that the right to obtain evidence from non-parties is sometimes necessary to ensure a fair trial. While the Supreme Court has not directly addressed scientific privilege, lower federal courts have applied a balancing test that weighs the requesting party’s need for the information against the burden that would be placed on the receiving party.[ii] When determining what information must be disclosed, courts have considered several factors, including the completeness of the information, confidentiality of sources, and chilling effects on research.[iii] Federal courts generally favor disclosure under the assumption that litigants have a right to review relevant evidence when resolving disputes.[iv] However, courts have made exceptions where there are public policy reasons for excluding evidence.[v] For example, courts have denied subpoena requests for materials that solely reveal researchers’ pre-publication deliberations unless there is a strongly compelling need by the requesting party.[vi] The outcome of the analysis often depends on the type of information being requested. The balancing test is a useful method to adjudicate the competing interests of each side when deciding whether to enforce a subpoena. However, the factors that affect the probative value of preliminary or deliberative scientific materials, and the burden placed on the researchers who produce them, are unique, and not all are thoroughly considered in the balancing test as applied by courts. Several courts hold that data from studies expected to be relied on during trial is highly probative.[vii] Indeed, this is in line with courts’ use of scientific findings as evidence in judicial proceedings. However, unpublished working materials are generally considered insufficiently probative to justify the burden of discovery on researchers. Unpublished working materials can be broadly divided into two categories: preliminary data and preliminary analyses, the latter referring to materials created as part of the deliberative process of interpreting the data and generating conclusions. In Deitchman v. E.R. Squibb & Sons, Inc., the Seventh Circuit states, “No discovery should be allowed of any material reflecting development of [the researcher’s] ideas or stating his or others’ conclusions not yet published.” 740 F.2d at 565. In United States v. Allen, the court found that the requested data was too preliminary and thus insufficiently probative to justify enforcing a subpoena for its disclosure.[viii] In Allen, the researchers had recently begun a study to measure the long-term toxicity of low doses of 2,3,7,8-tetrachlorodibenzo-p-dioxin (“TCDD”), a component of an herbicide that was the subject of an EPA hearing. The researchers protested the subpoena because the animal subjects had not yet received the full series of TCDD doses. The court held that the preliminary data available from the studies had very little probative value. 494 F. Supp. at 113. In Deitchman, a leading researcher had established and was maintaining a database recording adenocarcinoma cases that were thought to be linked to using the drug diethylstilbestrol (“DES”). Squibb obtained a subpoena that essentially requested every document in the database. The researcher challenged the subpoena, in part because it would force him to prematurely disclose his study results prior to peer review. The Deitchman court agreed with the researcher and limited discovery to factual materials from which Squib could perform its own analyses. 740 F.2d at 565.  In In re Bextra, the District Court of Massachusetts also held that peer review comments had only limited probative value. 249 F.R.D. at 12. These decisions together suggest that unpublished working materials should generally be considered non-probative. In limited circumstances, unpublished working materials may be considered sufficiently probative to justify discovery, particularly when the researchers’ behavior is at issue. If the party requesting the subpoena has a legitimate claim of potential fraud in the requested research, preliminary analyses may have higher probative value because they may reflect unethical or illegal behavior by the researchers. For example, fraud may prejudice a study’s results in a way not easily determinable from the study’s factual information. In Deitchman, the court notes that Squibb did not allege any impropriety in the research, suggesting that a colorable claim of fraud could affect a court’s determination of probity, even where the researcher is not party to the original suit. 740 F.2d at 563. There are public policy arguments in favor of broader disclosure where research fraud is suspected. Not all researchers behave ethically, and judicial investigation of allegations of scientific fraud would reduce the negative impact of false studies on the litigants and the public, whose taxes fund much of the research and whose safety may be affected. As just one example, a Massachusetts anesthesiologist published papers on research he never performed regarding commonly used painkillers.[ix] Discovery may also be appropriate in cases where a corporation, such as a drug manufacturer, is accused of hiding negative data from safety studies. However, general, unsupported allegations of fraud should not be used to justify abusive discovery to harass researchers who study politically sensitive subjects.[x] Balanced opposite the probative value is the burden placed on the subpoenaed. The burden on researchers when disclosing study-related materials is extremely high, especially when the studies are unpublished. Some courts have alleviated the financial burden on subpoenaed researchers by ordering the requesting party to pay for the cost of production, but less monetizable burdens remain.[xi] Some of these non-monetary burdens, such as the burden of disclosing proprietary knowledge, can be alleviated by protective orders against public disclosure, but courts have recognized that protective orders cannot adequately address all such burdens. The Allen court acknowledges that forcing the disclosure of preliminary information “is likely to jeopardize the study by exposing it to the criticism of those whose interests it may ultimately adversely affect, before there has been an opportunity for the researchers themselves to make sure the study is the result of their best efforts. This is not the kind of burden which can be lightened by a protective order.” 494 F.Supp. at 113. Similarly, other courts and commentators emphasize the reputational consequences that could result from disclosure of inaccurate preliminary data or conclusions, especially to a non-neutral party whose litigation goals directly oppose the researchers’ results. 672 F.2d at 1276. This is of special concern when the researchers are not parties or witnesses to the lawsuit, because they then have no opportunity to refute any impeachment arguments made against their work. Courts have also recognized the general chilling effects that forced disclosure produces on free academic research. Id. In the non-precedential WHOI order, the court excluded from discovery documents associated with academic studies published after the government report, agreeing that disclosure of those materials to BP could “hamper future research efforts.”[xii] Discovery of research materials is thus only warranted where the probative value is sufficiently large to justify imposing court-recognized burdens on the researchers. However, the burden of disclosure on researchers is greater than that described by the courts. Courts have focused on the harms to the researcher from potentially disclosing inaccurate data, but even disclosure of accurate, unpublished working materials is harmful because it damages researchers’ chances to publish. Researchers, especially academic researchers, exist in a “publish or perish” environment, and the success of their careers depends in large part on the number of papers they publish in high-profile journals. Disclosure of working materials burdens researchers’ ability to publish in two ways. First, academic journals value previously unknown findings. Early disclosure of a study destroys the novelty of a later-submitted manuscript, and thereby drastically reduces the researchers’ chance of having their manuscript accepted. Courts could address this problem by issuing protective orders that place study-related information under seal prior to publication — although leaks are still possible. Second, and perhaps more importantly, is that researchers must spend time gathering the information. Regardless of how broad or narrow the subpoena request, researchers will have to comb through lab notebooks, databases, email records, and other materials. Given the specialized nature of research, it would be difficult if not impossible for a researcher to hire an outsider to gather the information, because no one else could identify the materials relevant to a particular study. Thus, the researcher (or his graduate students) will be forced to gather the information themselves. All this time is time not spent on the researchers’ work. In highly competitive fields, several labs may be competing to complete and publish the same discovery. Even a few days lost at a critical time can severely damage a researcher’s progress and result in a competitor “scooping” their research. Such reputational burdens cannot be alleviated by protective orders or financial compensation. These reputational burdens affect academic researchers more strongly than corporate researchers. Success in industry research depends on successful product development, not on publication. Though publishing peer-reviewed articles does play a role in companies’ marketing strategies, corporations often decline to seek publication of their studies, which suggests that the pressure to publish is not as severe. There is still some burden on corporations and corporate researchers, especially with regard to the competitive disadvantage that results from being required to spend time on disclosure. However, while first-mover and first-to-file patent advantages are important to corporations, these factors are less immediately dependent on research progress than publication is, so the burden on disclosure is similarly smaller. Thus, a court can generally weigh the publication burden more heavily for academic compared to corporate research. One concern with this distinction is that there is an increasing trend of corporate sponsorship of research at academic institutions, which sometimes blurs the line between academic and corporate research. A court may be able to resolve such situations by considering whether the researchers’ publication records strongly affect their career advancement or ability to obtain research funding. In some cases, such as where the requesting party has alleged scientific fraud, researchers’ work product should be discoverable. Only in such a narrow set of circumstances does the probative value of such materials outweigh the burden to the researcher. In most situations, working materials are not strongly probative, and the burden of disclosure is unjustifiably high. Therefore, courts that apply the balancing test to subpoenas of non-party researchers should generally limit discovery of unpublished working materials.
[i] Order Regarding BP’s Subpoena for Woods Hole’s Analysis Documents, Case 2:10-md-02179-CJB-SS (E.D. La. 2012).
[ii] See, e.g., Anker v. G.D. Searle & Co., 126 F.R.D 515, 519 (M.D. NC 1989). Courts have consistently held that the researcher cannot be excused from disclosure solely on the basis of being a non-party and a non-witness.
[iii] See, e.g., Cusumano v. Microsoft, 162 F.3d 708, 717 (1st Cir. 1998); United States v. Allen, 494 F. Supp. 107, 113 (W.D. Wis. 1980). aff’d, Dow Chemical Co. v. Allen, 672 F.2d 1262 (7th Cir. 1982). [iv] See, e,g., Wright v. Jeep Corp., 547 F. Supp. 871, 874 (E.D. Mich. 1982).
[v]  Id. at 873.
[vi] See, e.g., Allen, 494 F. Supp at 113; In re Bextra and Celebrex Mktng. Sales Practices and Prod. Liab. Litig., 249 F.R.D. 8, 12 (D. Mass. 2008).
[vii] See, e.g., In re American Tobacco Co., 880 F.2d 1520, 1529 (2d Cir. 1989); Wright, 547 F. Supp at 874.
[viii] Allen, 494 F. Supp at 113. But see Deichtman v. E.R. Squibb & Sons, Inc., 740 F.2d 556, 565 (7th Cir. 1984) (holding that the researcher’s unique preeminence in the field and the expert witness’s expected heavy reliance on his studies justified requiring disclosure of preliminary data). This case suggests that preliminary data may be less protected than materials reflecting ongoing development of ideas or conclusions. [ix] Scott S. Reuben: Debarment Order, 76 Fed. Reg. 71,042 (Food & Drug Admin. Nov. 16, 2011).
[x] Rector and Visitors of University of Virginia v. Cuccinelli, 80 Va. Cir. 657, 658 (2010) (holding that the Attorney General of Virginia failed to allege any conduct that violated a statute in its Civil Investigative Demands, which were aimed at forcing the University of Virginia to disclose data and other materials related to a climate researcher’s grant applications when employed at the university), aff’d on other grounds.
[xi] One major category of data, confidential information obtained from human research subjects, is not addressed here. Courts have generally found that there is high presumption against discovery of private information given in confidence to researchers, similar to the deference given to journalists’ confidential sources. See, e.g., Cusumano, 162 F.3d at 714.
[xii] Order Regarding BP’s Subpoena for Woods Hole’s Analysis Documents, Case 2:10-md-02179-CJB-SS at 10 (E.D. La. 2012).