## Facts

- Hundreds of plaintiffs in a federal products-liability MDL alleged injuries from Biomet’s metal-on-metal hip implant system.
- The dispute concerned management of electronically stored information (ESI) discovery, not the merits of liability or causation.
- Biomet identified an initial ESI universe of about 19.5 million documents potentially responsive to discovery requests.
- Using negotiated keyword searches, Biomet reduced that universe to a pool of about 2 million documents for further review.
- Biomet proposed applying predictive coding (technology-assisted review) to the 2 million documents, supported by statistical sampling indicating high confidence in locating responsive material.
- Plaintiffs’ steering committee sought an order requiring Biomet to apply predictive coding to the entire 19-million-document universe, arguing keyword culling could have excluded relevant documents and thereby “tainted” the review set.
- Biomet argued the requested redo would impose substantial additional cost (potentially millions of dollars) and delay, with uncertain marginal benefit, after significant expenditures already incurred.

## Issues

1. Under Federal Rule of Civil Procedure 26, should Biomet be compelled to run predictive coding against the entire 19-million-document ESI universe rather than using predictive coding on the 2-million-document set produced by prior keyword culling?
2. Does the asserted imperfection of keyword searches require restarting the ESI workflow when the producing party’s process is otherwise supported by sampling and proportionality considerations?

## Decision

- The court denied the plaintiffs’ request to compel predictive coding across the entire 19-million-document universe.
- The court permitted Biomet to proceed with a hybrid workflow: keyword culling followed by predictive coding on the reduced set, with sampling and quality-control measures.
- The court found the plaintiffs had not shown that the likely incremental benefit of restarting discovery outweighed the added burden, expense, and delay.

## Legal Principles

- Discovery search methodology is governed by Rule 26 reasonableness and proportionality, not a requirement to use any specific technology or sequence.
- Rule 26(b)(2)(C) authorizes limiting discovery when the burden or expense outweighs the likely benefit, considering the needs of the case and the importance of the discovery to resolving issues.
- A requesting party generally cannot force a producing party to adopt the requester’s preferred ESI search method when the producing party’s approach is defensible, validated, and proportionate.
- Imperfect recall is not, by itself, a basis to compel substantially more expensive discovery; statistical sampling and quality control can support a finding that an ESI workflow is reasonable.

## Conclusion

The court held that, in MDL-scale ESI discovery, proportionality under Rule 26 permits a producing party to use a reasonable hybrid approach (keyword culling followed by predictive coding) and does not require restarting with predictive coding over an entire corpus absent a concrete showing of material benefit that justifies the added cost and delay.


In re Biomet M2a Magnum Hip Implant Prods. Liab. Litig., No. 3:12-MD-02391, 2013 WL 1729682 (N.D. Ind. Apr. 18, 2013)