A new case has brought into the spotlight the risks responding parties face when entering into ESI protocols with detailed disclosure obligations regarding technology-assisted review (“TAR”). In In re Valsartan, Losartan, and Irbesartan Products Liability Litigation (D.N.J. Dec. 2, 2020), the court refused to approve a responding party’s proposed TAR process and insisted that the design and execution of a TAR workflow required “an unprecedented degree of transparency” between the responding and requesting parties.
Valsartan’s holding flows from the stipulated ESI protocol in which the parties agreed to “cooperate” and “meet and confer” regarding “the disclosure and formulation” of TAR. Indeed, the Valsartan holding is specifically premised on the ESI protocol’s expansive transparency and cooperation provisions, to which a responding party agreed and then failed to observe. Given Valsartan’s holding and rationale, responding parties may be wary of entering into ESI protocols with such broad disclosure and cooperation obligations regarding TAR.
The Impact of the ESI Protocol on TAR
Valsartan is an MDL matter in which plaintiffs are seeking relief for alleged health problems arising from their use of prescription medication to treat high blood pressure. To better handle electronic discovery in a sprawling MDL, the parties stipulated to an ESI protocol that would ostensibly address various issues, including search methodologies. For example, all parties agreed they would “cooperate in good faith regarding the disclosure and formulation of appropriate search methodology.” This provision applied to both search terms and TAR. In addition, the parties agreed they would undergo a fulsome disclosure and formulation process “prior to using” TAR and that they would also “meet and confer as early as possible to discuss” applicable search methodologies including TAR. These are the provisions that caused so much trouble for one defendant, Teva Pharmaceuticals (“Teva”), who sought to use TAR.
Disputes over Search Terms and TAR
After a year of protracted negotiations over multiple sets of search terms—complicated by the failure of Teva (and the other defendants) to sample and test the initial set of terms—the parties stipulated to a search term protocol, which the court then entered as an order.
Less than a week later, Teva disclosed to plaintiffs for the first time that it would be using a TAR 2.0 (continuous active learning) process to identify and produce relevant information. Plaintiffs objected, asserting Teva should not be permitted to use a workflow that leveraged both search terms and TAR.
Despite their dispute, the parties nearly executed a protocol that would have cemented with the court’s approval Teva’s use of TAR. Nevertheless, Teva did not execute the separate TAR protocol because it would not agree to a provision requiring it to disclose to plaintiffs 5,000 nonresponsive documents for their review as part of the validation process.
As a result, Teva informed plaintiffs and the court that it would not use TAR to eliminate non-responsive documents. Instead, Teva represented that it would manually review all of its documents despite anticipating the “extremely burdensome” nature of such a process.
Notwithstanding that representation, Teva proceeded to use CMML TAR 2.0 software for its review. Teva revealed this for the first time two months later when it returned to court asserting that during the intervening time, it had developed TAR validation measures which, it argued, confirmed the nonresponsive nature of hundreds of thousands of documents. In reliance on its unilaterally designed validation measures, Teva asked the court to approve its TAR process and foreclose on proportionality grounds any further review of the predicted nonresponsive documents. Alternatively, Teva requested the court to shift the costs of the manual review to plaintiffs.
The ESI Protocol Governs the Disposition of the Parties’ TAR Dispute
In response, the court issued a scathing opinion, rejecting Teva’s requested relief and criticizing Teva’s conduct in connection with the search term and TAR negotiations. The court did not take issue with the merits of Teva’s TAR process. For example, the court rejected plaintiffs’ argument that Teva could not use search terms in connection with TAR to help identify relevant information. The court also acknowledged the application of Sedona Principle Six, i.e., “that a producing party has the right in the first instance to decide how it will produce its documents.” Nevertheless, the court held that Teva could not rely on Sedona Principle Six to circumvent its transparency and cooperation duties to which Teva voluntarily agreed when it executed the ESI protocol.
Against this backdrop, the court held that Teva should have disclosed its intention to use TAR long before the parties finalized their search term protocol. Because the parties “should have known early on that ESI discovery was going to be costly,” the court reasoned that Teva should have considered the use of TAR “at the outset of the litigation.”
The court also faulted Teva for failing to involve plaintiffs in the design of Teva’s TAR workflow under the “meet and confer” and “formulation” provisions in the ESI protocol. The unilateral manner in which Teva prepared its TAR process violated the protocol and prevented plaintiffs from “meaningfully contribut[ing]” to the TAR process. All of which led to the court’s conclusion that Teva violated the ESI protocol.
Despite its frustration, the court did not obligate Teva to manually review the predicted nonresponsive documents given the cost of such a process (estimated to be in excess of $1,000,000). Instead, the court ordered Teva to review those documents using TAR pursuant to the protocol the parties previously negotiated but from which Teva had walked away.
Over Teva’s renewed objections, that protocol would include the disputed validation provision allowing plaintiffs to review 5,000 nonresponsive documents “of their choosing.” In doing so, the court dismissed Teva’s concerns that its order was tantamount to a production of irrelevant documents. The court went so far as to express surprise at Teva’s reluctance to produce predicted nonresponsive documents, conjecturing that it “does not understand why this provision is so bothersome to Teva.” The court closed by declaring that its opinion should be “the last word” on any further disputes between the parties over TAR.
Valsartan’s Impact on ESI Protocols
There are any number of instructive lessons from Valsartan. To be sure, Valsartan teaches the importance of observing court-ordered ESI protocols, considering search and review options at the outset of litigation, and timely seeking judicial relief from onerous discovery obligations.
But the key point from Valsartan is undoubtedly its impact on ESI protocols. Given the court’s sweeping statements about transparency and cooperation when using TAR pursuant to an ESI protocol, responding parties could very well be reluctant to enter into such protocols or even disclose the use of TAR. Doing so could leave responding parties vulnerable to a weakened TAR process after judicial retrospection just as it did to Teva. Indeed, Teva is now being forced to disclose thousands of irrelevant documents and do so arbitrarily without any sampling or custodial specifications.
That Valsartan may result in fewer ESI protocols addressing TAR or other search issues is a lamentable development given the advantages that adversarial cooperation and reasonable transparency offer to parties in discovery. Valsartan hopefully will not be “the final word” on these issues.