Takeda Pharmaceuticals faces multiple lawsuits alleging that it was aware of and willfully concealed a link between one of its most popular drugs, Actos, and increased risk for bladder cancer. This case is particularly interesting because of its potential to set guidelines for the future use of Technology Assisted Review (TAR) in discovery. An order written by Western District of Louisiana Judge Doherty provides specific instructions to both parties for conducting their discovery of Electronically Stored Information (ESI) using TAR.

Judge Doherty’s order continues recent precedents in ESI standards in that it encourages a great deal of cooperation between parties during the discovery phase of litigation. Notably, if the parties disagree about the privilege status of any documents they will be required to meet and confer prior to contacting the court for resolution.

The order provides a framework for both parties to conduct discovery. Of particular note is the guidance that Judge Doherty provides for the Assessment Phase of e-discovery. Following this framework, the timeline of the assessment phase for the case is as follows:

1. Both parties will meet and confer to determine the names of four key Takeda custodians from which to pull a randomized sample of 500 documents.
2. Common irrelevant documents such as Spam, Commercial e-mail, System files, etc. will be removed from the sample population
3. Takeda and Plaintiffs will each nominate three experts to work collaboratively to train the TAR software. Plaintiff’s experts must agree not to disclose any information subject to withholding or redaction to co-counsel, client, any Party, or any third party without prior written consent from opposing counsel.
4. Takeda’s experts will review documents for privilege prior to Plaintiffs’ experts viewing them. These will be withheld from viewing by Plaintiff’s experts or redacted, and a privilege log will be provided. These privileged documents can still be used to train the software.
5. Experts from both parties will work collaboratively to determine the relevance of non-privileged and privileged-redacted documents. Takeda’s experts will determine the relevance of privileged-withheld documents.
6. The initial Control Set of 500 documents uses a 95% confidence level for estimating richness (the ratio of relevant to irrelevant documents). There is a worse-case error margin of plus or minus 4.3% assuming richness of 50% and would lower the richness of the set. For the software to function effectively, the Control Set must reach the “Statistical” level of validation, meaning it contains at least 70 relevant documents. For a document set with 14% richness and above, a Control Set of 500 documents is sufficient. For lower levels of richness, experts must review additional documents in the Assessment Phase.
7. In order to minimize the margin of error on the TAR software’s estimates, both parties will continue reviewing documents in the Assessment Phase beyond the “Statistical” level until the Control Set contains at least 385 relevant documents.

As Technology Assisted Review becomes more and more prevalent, guidelines like these will become increasingly important. This order sets realistic terms for the use of TAR that could help form the foundation for rules governing its use in e-discovery.