Keyword Filtering Prior To Predictive Coding Deemed Reasonable

In re: Biomet M2a Magnum Hip Implant Prods. Liab. Litig., NO. 3:12-MD-2391 (N.D. Ind. Apr. 18, 2013)

The United States District Court for the Northern District of Indiana recently recognized that an approach to eDiscovery that incorporates keyword searching as an inital culling tool and predictive coding as a secondary cull can be reasonable in even the largest of cases.  Biomet was being sued by hundreds of individual litigants and classes over its product.  Before the Judicial Panel on Multidistrict Litigation decided to centralize the litigation, Biomet took an anticipatory approach to discovery, over the objections of plaintiffs’ counsel who advised them to wait.  First, Biomet identified a universe of 19.5 million potentially relevant documents.  It then used keyword culling to trim this down to 3.9 million, with de-duplication further reducing the field to 2.5 million documents.  Statistical sampling of the excluded documents verified that less than 2.5% were potentially relevant.  Biomet then used predictive coding to rank the documents and contract attorneys to review for relevance, confidentially and privilege.  The cost of this was over $1,000,000.

The Plaintiffs were unsatisfied.  They argued that predictive coding should have been applied to the original universe of 19.5 million documents, and that using keyword searches was less effective and “tainted” the resulting cull.  They wanted the court to order Biomet to start from scratch, which would have cost Biomet at least as much, if not more, than it had already spent.

The Court declined to so order.  After noting that Plaintiffs’ arguments were supported by little more than a New York Law Journal article lacking in sources, the Court refused to address whether predictive coding is “better” than keyword searches.  The Court instead noted that the question was whether Biomet had done enough to meet its discovery obligations under Federal Rule of Civil Procedure 26(b) and 34(b)(2).  The Court found that it had.  Even given the large amount in controversy, the parties’ resources, the important issues at stake and the importance of discovery in resolving them, the Court found that the likely benefit from pursuing Plaintiffs’ proposed approach did not equal or outweigh the additional burden and expense on Biomet.

The Court was no doubt influenced by Biomet’s anticipatory approach and its own efforts to contain costs, which it viewed as reasonable.   Elijah has developed a predictive coding solution, ESPI, that marries the best of both worlds – allowing attorneys to leverage the use of keyword searches while obtaining the benefits of predictive analytics in substantially reducing the burdens associated with document intensive reviews.  For additional information about Elijah’s ESPI predictive coding solution, please visit