In re: VIOXX PRODUCTS LIABILITY LITIGATION Steering Committee Plaintiff-Appellee v. Merck and Company Inc Defendant-Appellant. In re: Merck and Company Inc Petitioner No. 06-30378, 06-30379 United States Court of Appeals, Fifth Circuit May 26, 2006 Appeal from the United States District Court for the Eastern District of Louisiana. USDC No. 2:05-MD-1657. Petition for Writ of Mandamus to the United States District Court for the Eastern District of Louisiana, New Orleans. Counsel Russ M. Herman, Herman, Herman, Katz & Cotlar, New Orleans, LA, for Plaintiff-Appellee. Phillip A. Wittmann, Dorothy Hudson Wimberly, Stone, Pigman, Walther & Wittmann, New Orleans, LA, Douglas Marvin, Williams & Connolly, John H. Beisner, O'Melveny & Myers, Washington, DC, for Defendant-Appellant, Petitioner. Per Curiam Opinion Before HIGGINBOTHAM, DENNIS, and OWEN, Circuit Judges. *1 Merck Company, Inc. seeks review by appeal and mandamus of the district court's ruling on its claim of attorney-client privilege for thousands of documents sought by plaintiffs in the Vioxx Products Liability Litigation now pending as an MDL assignment in the Eastern District of Louisiana before Judge Eldon Fallon. As we will detail, mandamus is the only avenue of review.[1] I We are told that Merck has produced over 2,365,000 documents, amounting to approximately 18 million pages, all in response to discovery requests in this litigation. While Merck claims privilege for only one percent of that number, the privileged document collection reviewed by the district court contains some 30,000 documents. The MDL proceeding, established in 2005 to coordinate pretrial proceedings for personal injury and consumer fraud cases, now includes approximately 5,200 individual personal injury cases and numerous class actions. Resolution of the claim of privilege for 30,000 or so of the 2,365,000 documents already provided-in less than a year-is crucial to these many cases. The plaintiffs in this MDL proceeding are before us as the plaintiff steering committee (PSC). The PSC challenged Merck's privilege designations on October 7, 2005, urging that certain categories of documents enjoy no privilege and that the privilege log was not sufficiently descriptive. The PSC also observed: “Obviously, a document-by-document review is both impractical and inefficient. Instead, a feasible alternative is an in camera review of samples of documents from each of several categories identified herein.” II On November 7, 2005 the district court ordered Merck to submit within ten days, under seal, all documents in which privilege was claimed for review. Merck provided the court with the documents, its privilege log, and offered legal personnel and outside counsel to assist.[2] On February 2, 2006 the court ordered Merck to divide the documents into agreed-upon categories, such as marketing documents, regulatory documents, etc., this to enable the court to examine “random samples of each category” and render an order as to Merck's claim of privilege. On April 3, the district court noted by minute order that it had removed all documents from boxes 1-20 for which it sustained the privilege and ordered the lawyers to confer on the method for the review by the PLC of the remaining documents. A similar order followed on April 4 for boxes 21-30 of Merck's “privileged” documents. Two days later the court held a status conference with counsel at which Merck advised that it wanted to file a motion for reconsideration of “a portion of the documents already reviewed by the court.” The court ordered that the motion must be filed the next day with any reply the following Monday, April 10, 2006. In its motion, Merck reminded the court that in October it had been urged to conduct “an in camera review of a sample of documents listed on the privilege log and apply privilege determinations categorically to similar documents ...” and that the court, presumably experiencing difficulty with the random selection from categories protocol, instead began a document-by-document review of several boxes of documents. Merck explained that because it was uncertain of the process that the court intended to use in deciding the privilege question, Merck did not “... at that time provide ... additional contextual information” but stood ready to do so for any document needing it. Merck urged the court to consider the totality of the circumstances surrounding each document, pointing to In re Grand Jury Proceeding.[3] *2 On April 19, the district court heard argument on the motion for reconsideration and denied it, offering his reason from the bench. In sum, the judge described his frustration in reviewing the documents, with a privilege log of a thousand pages in length and eighty boxes of documents. The district court did not rule on the claimed inadequacy of Merck's privilege log. He explained that for many documents Merck failed to maintain adequate entries in the privilege log, that the attempt to use categories of documents proved impracticable, and that he commenced his seriatim examination. Consequently, after examining approximately 30,000 documents, the district court upheld the privilege claim for approximately 491 documents. It did not request any additional assistance from Merck. The district court refused to stay its discovery order. We issued a stay of the district court's order to gain sufficient time for appellate review. Sensitive to the time demands of the litigation, on receipt of briefs, we held oral argument by telephone conference call. III Today, we examine only the process, not the merit of the privilege claim on any single or group of documents. The district judge undertook the herculean task of personally reviewing 30,000 documents over a two-week period. Unfortunately, this effort produced over 200 documents in which it reached inconsistent determinations, concluding that one copy of a document was privileged and that exact duplicates of the same document were not. Despite the commendable effort of this able district judge, we must conclude that the process that has evolved in the litigation has proved to be inadequate and to continue on this course would be an abuse of discretion.[4] That said, we limit our response to suggesting a slight change of course. In oral argument Merck expressed willingness to identify 2,000 documents for further examination by the district court or its designee. The proposal includes the offer of counsel at hand to answer questions of the examining official or to provide affidavits under seal when requested. This process is offered as a means by which the “more significant documents” may be examined with full assistance to the district court, producing a pattern that may validate the earlier process and facilitate any further proceedings. Plaintiffs want to end the effort, arguing that Merck has failed in its obligations under Rule 26 to provide an adequate log and that it has effectively waived any claim of attorney-client privilege. We are not persuaded that Merck has waived any attorney-client privilege it may have with the documents at issue. As we see it, the claim of waiver by the PSC seeks too much. We are keenly aware that this MDL litigation is comprised of many cases that are ready for trial, as well as the larger numbers that are not, and that the district court in its impressive management effort is wisely planning to capture the benefits of completed trials. All seek prompt rulings. And do we. Plaintiffs have persuaded the district court to undertake in camera review. Given the error rate of the various courses taken to date, the resulting wholesale rejection of privilege now ordered cannot stand. At the same time, turning to the proposed examination in camera of 2,000 documents, it is apparent that a new protocol is needed. The district court must be provided the assistance it should have had all along-the opportunity to review documents against the log as supplemented on its request, whether each document is examined or the examination is a random sampling from the universe of 2,000.[5] *3 Such assistance is a practical necessity in the real world of MDL discovery. Plaintiffs have the benefit of an in camera examination and Merck should have the opportunity to support its claim of privilege when it is necessary to do so. And the court must be given the support it needs. Merck, as we understand it, will place personnel at hand to answer questions of Judge Fallon or his designee and provide such assistance as he may need, including arranging the documents in ways that will ease the burden of review with or without such randomization as the trial court may order. This supplementation will include written filings if requested. Because this process is by necessity ex parte, all oral communications must be recorded by a court reporter at Merck's expense, with all transcripts and written submissions filed under seal, a necessary discipline to the process.[6] Merck's proposal is a sound solution and must be pursued. We accept that all parties will tack to this course. We decline today to issue any writ or attempt any further ruling. Specifically, we do not decide the extent of any further document examination beyond the 2,000 documents Merck would put at issue. The district court will have the benefit of that effort and is best situated to make that decision. In short, whether examination in camera should extend beyond the 2,000 documents will reside in his good judgment. We understand that Merck will produce the identified 2,000 documents not later than the first business day ten days following the filing of this order and stand ready to assist the district court, as we have described. We leave to the district court whether it should undertake this effort itself or seek the assistance of a special master or magistrate judge. With respect, we are not persuaded that the record supports the dissent's characterization of it, drawing more than we on the rhetoric of PSC's briefs. We do not demean the work of the district court in this difficult matter. Consolidated discovery of MDL litigation, with its huge aggregation of cases, sorely taxes the processes attending our traditional binary structure in civil cases. Consolidated discovery offers efficiencies, but at a price. These massive consolidations have a distorting effect on pretrial procedures, stretch attorney-client relationships, and re-allocate economic risk. Our point is not about the merits of the many trade-offs. It is rather that there are limits to the adaptability of our trial processes, flexible as they are. The problem at bottom in this case is the large number of documents demanded and produced with the attending claims of privilege. Traditional procedural protections are not limitlessly malleable. If staying within those traditional constraints takes more time than jumping their traces, that is not justification for doing so. The time it takes is the time it takes. It is so ordered. DENNIS, Circuit Judge, concurring in part and dissenting in part. DENNIS, Circuit Judge. *4 I concur in the panel majority's decision to decline to issue any writ at this time, and its ruling that we lack jurisdiction to hear Merck's appeal, but I respectfully dissent from its failure to deny the petitioner's motions for mandamus and emergency stay, its failure to dismiss the appeal and to either dismiss this panel's own stay or else amend it to make it subject to the order of the district court. Judge Eldon E. Fallon is coordinating the pretrial discovery proceedings in the Vioxx Multi-District Litigation that commenced in 2005 consisting of some 5,200 personal injury and consumer fraud cases and class actions. In addition, Judge Fallon is conducting the trials in a number of the Vioxx cases. He has already completed two Vioxx trials and has four others set for July, August, September and October or November of 2006. This discovery controversy arises out of the Vioxx case set for trial before Judge Fallon in July, 2006. The record before us consists of counsel's written arguments, unsworn exhibits, unrecorded oral telephonic arguments, and transcribed oral arguments and Judge Fallon's oral reasons in the district court. The Plaintiffs' Steering Committee (PSC) sought discovery of a large number of documents from the Defendant-petitioner, Merck & Co., Inc. (Merck), of which Merck withheld some from full disclosure, asserting attorney-client privilege. After conducting an in camera examination of the 31,846 documents in question, 16,851 withheld completely and 14,995 containing redactions, Judge Fallon upheld Merck's privilege claim as to 491 documents and ordered the remainder disclosed as unprivileged. Merck moves this court for mandamus and an emergency stay and filed an interlocutory appeal, seeking this court's judgment requiring Judge Fallon to defend his order against Merck's challenges of “representative documents” and/or to allow Merck to provide additional support for its privilege claims. The PSC contends that it began in July, 2005 to press Merck for a privilege log in compliance with FED. R. CIV. P. 26(b)(5) but that Merck continually refused to comply or adhere to court orders, “failed to adequately supplement its log with sufficient descriptions, and ‘de-privileged’ and produced previously withheld documents within weeks of trial” in other Vioxx cases. The PSC describes in detail seven occasions between July 2005 and February 15, 2006, on which it and/or the district court pressed Merck for correction of the vagueness and inadequacy of the privilege log. Merck, on the other hand contends that it has already produced over 2,365,000 documents and that its privilege log, after several amendments, is adequate. The PSC avers that Merck purposefully delayed until November 4, 2005, in releasing 220,000 pages of its unprivileged documents that it was required to disclose in July 2005, and similarly de-privileged and released thousands more on January 30, 2006, thereby prejudicially depriving plaintiffs in this and other Vioxx cases of adequate discovery and trial preparation. Merck concedes that its privilege claim may have appeared to be aggressive to the district court, but it argues that privileged designations can often be close calls; that it has never attempted to hide or delay the release of unprivileged material; and with reference to the 220,000 documents it released belatedly in November 2005, its counsel told the district court: *5 When you're producing millions and millions of documents, your honor, you sometimes make mistakes. We concede that that has happened here. But we certainly have done our best to try and fulfill our obligations to the court and to opposing counsel. Transcript of motion to stay proceedings, April 24, 2006, at 22. Judge Fallon orally explained his procedure in conducting the in camera examination and stated the reasons for his rulings in the district court hearings. When the parties were unable to agree upon a privilege log with sufficient clarity and specificity, Judge Fallon ordered that the contested documents be given to the court in camera. Eighty-one boxes were delivered, consisting of five to six thousand documents per box, containing a total of nearly 500,000 pages. With the parties' agreement, he planned to have the documents divided into categories, take a random sampling from each category and declare each whole category either privileged or unprivileged based on his sample findings. He instructed Merck to provide him with a privilege log with the categories fleshed out. When Judge Fallon received the 81 boxes of documents, he soon discovered that the procedure he had agreed upon with the parties was impracticable. First, the documents had not been arranged physically into categories within the boxes. Merck had simply given him an electronic categorization. Documents from the same category were strewn throughout all or many of the 81 boxes. Thus, before he could begin to sample a percentage of each category the relevant documents needed to be assembled manually. This would have required too much time and duplication of effort. Second, an extremely high percentage of Merck's documents showed that in-house attorneys had been copied on them and there was no way to quickly and easily determine whether copies of a particular document had been sent to in-house counsel with respect to facilitating or soliciting legal advice or for a non-germane purpose. Third, many of the documents contained anonymous handwritten notes, which could not easily or efficiently be attributed to attorneys, non-attorneys, or related to or disassociated from attorney-client communications. Fourth, Judge Fallon stated that the privilege log and the documents inadequately identified and distinguished attorneys from non-attorneys. Merck disputes this and claims that it “provided the court with a list ‘that identifies lawyers, paralegals and administrative assistants to lawyers whose names appear in the privilege log.” ’ ... “On the privilege log itself, Merck also marked all lawyers and legal personnel with an asterisk.” But ambiguity and confusion is self-evident in Merck's identification and labeling scheme. “Paralegals,” and “administrative assistants to lawyers” are not lawyers, and “legal personnel” may include more than just lawyers. Additionally, persons with law degrees do not necessarily act as lawyers in respect to all of their communications, actions and advice. See WRIGHT & MILLER, FEDERAL PRACTICE AND PROCEDURE §§ 5480-5486. Fifth, Judge Fallon found that the documents often did not fit snugly within the designated categories. That is, “it was very difficult to put them in categories because they were different, each document to some extent is different.” Sixth, ambiguity and confusion plagued many of the documents because of the description of their subject matter as “reflecting legal advice” rather than stating specifically if senders or receivers of the documents gave or received legal advice. Thus, a document examiner could experience difficulty in telling whether the communicators had dispensed or received legal advice. *6 Therefore, Judge Fallon began to examine the documents on a document-by-document basis. After he had examined the documents in 30 of the boxes in this manner, he notified Merck that he was forced to abandon the plan of dividing and sampling the documents categorically. At that point, he removed the privileged documents from the first 30 boxes, and made them available to Merck so that it could see which documents he had designated as unprivileged. He invited input from Merck and immediately received its objection and some additional data helpful in identifying communicators and legal advice. Merck filed a motion for reconsideration, urging the court to “establish a procedure to allow Merck to defend its privilege designations ... and reconsider its process for examining the remainder of the documents at issue.” It submitted 25 “representative” sample documents with explanations of why it felt each should be privileged. Judge Fallon reviewed the remaining 51 boxes using this data and then reviewed the first 30 boxes again using the additional data. With the help of the additional data he also reclassified five of 25 documents from the first 30 boxes as privileged which Merck had submitted as examples of improper privilege calls. He explained his privilege determinations as follows: I felt that the documents that were not privileged, were not privileged because either the privilege was waived, had no adequate privilege log, or that the documents had already been disclosed either during TV presentations or the letters were out in public. I had seen them during the course of the trial, and they were in some form or fashion no longer privileged, if ever they were privileged. In short, I felt the party urging the privilege failed to carry the burden. He further noted that he was surprised, given the seriousness of a claim of privilege, to find in the documents that were claimed privileged things such as promotional overviews, press releases, studies which are already into evidence in various forms, sales meetings, questions and answers regarding various TV interviews, statistical analysis protocol, discussion points for meetings. Before this panel, Merck argues that it is entitled to a writ of mandamus requiring the district court to defend its rulings against Merck's challenges in respect to “representative documents and/or to [allow Merck to] provide additional support for the privilege before the de-designation of privileged documents.” The grounds that Merck sets forth are that (1) the district court mistakenly ordered the release of 200 documents as not privileged although the court determined that exact duplicates of them were privileged; (2) the district court deemed certain attorney-client communications non-privileged because it could not discern which parties to the communication were attorneys-even though Merck's privilege log designated all attorneys with an asterisk and Merck provided the court with a list of attorneys listed in the log; (3) the district court's in camera review should be presumed to be invalid because it is impossible for Judge Fallon to have made consistent, sound privilege rulings on 30,000 documents within a two week period; and (4) the court erroneously rejected privilege claims regarding emails and other communications between Merck employees and the company's in-house and outside counsel that fall squarely within the attorney-client privilege. *7 Before considering whether any or a combination of these grounds is sufficient to warrant the extraordinary relief of mandamus, there is an inherent weakness in Merck's contentions that should be pointed out. In asking to adversely challenge the district court's some 30,000 rulings, Merck itself concedes that “the volume of documents in a proceeding as large as this one may make an adversarial exchange regarding the privilege's applicability to each document unfeasible[.]” There can be little doubt about this. Merck tacitly admitted this when it agreed to have the district court examine and base its determinations on its own in-camera samples of categories of the documents without any additional in-put from Merck. The panel majority obviously agrees. It requires the district court to reexamine only 2,000 documents and allows Merck to merely answer questions or provide affidavits when requested. I do not agree that there is any legal basis for this court to mandamus the district court to reexamine any of the documents, but I heartily agree that Merck is not entitled to challenge any and all of the district court's rulings it chooses or to put on additional evidence and argument not presented in the initial discovery proceedings. The Supreme Court has held that the remedy of mandamus is a drastic one, to be invoked only in extraordinary situations. Kerr v. U.S. Dist. Court for Northern Dist. of California, 426 U.S. 394, 96 S.Ct. 2119, 48 L.Ed.2d 725 (1976) (citing Will v. United States, 389 U.S. 90, 95, 88 S.Ct. 269, 19 L.Ed.2d 305 (1967); Bankers Life & Cas. Co. v. Holland, 346 U.S. 379, 382-385, 74 S.Ct. 145, 98 L.Ed. 106 (1953); Ex parte Fahey, 332 U.S. 258, 259, 67 S.Ct. 1558, 91 L.Ed. 2041 (1947)). The Court has observed that the writ “has traditionally been used in the federal courts only ‘to confine an inferior court to a lawful exercise of its prescribed jurisdiction or to compel it to exercise its authority when it is its duty to do so.” ’ Will, 389 U.S. at 95 (quoting Roche v. Evaporated Milk Assn., 319 U.S. 21, 26, 63 S.Ct. 938, 87 L.Ed. 1185 (1943)). And, while the courts have not limited the use of mandamus by an unduly narrow and technical understanding of what constitutes a matter of “jurisdiction,” Will 389 U.S., at 95, still “only exceptional circumstances amounting to a judicial ‘usurpation of power’ will justify the invocation of this extraordinary remedy.” Will 389 U.S., at 95. Such treatment of mandamus within the federal court system is founded on good reason. As the Supreme Court has repeatedly recognized, mandamus actions such as the one involved in the instant case “have the unfortunate consequence of making the (district court) judge a litigant, obliged to obtain personal counsel or to leave his defense to one of the litigants (appearing) before him” in the underlying case. Bankers Life & Cas. Co. v. Holland, 346 U.S. at 384-385 (quoting Ex parte Fahey, supra, 332 U.S. at 260). “More importantly, particularly in an era of excessively crowded lower court dockets, it is in the interest of the fair and prompt administration of justice to discourage piecemeal litigation. It has been Congress' determination since the Judiciary Act of 1789 that as a general rule ‘appellate review should be postponed ... until after final judgment has been rendered by the trial court.” ’ Kerr, 426 U.S. at 403 (quoting Will, 389 U.S. at 96 and citing Parr v. United States, 351 U.S. 513, 520-521, 76 S.Ct. 912, 100 L.Ed. 1377 (1956)). Issuing mandamus in “anything less than an extraordinary situation” risks defeating those very policies. Id. *8 Accordingly, the Supreme Court has set forth particular conditions for issuance of mandamus. Among these are that the party seeking issuance of the writ have no other adequate means to attain the relief he desires, Roche v. Evaporated Milk Assn., supra, 319 U.S., at 26, 63 S.Ct., at 941, and that he satisfy “the burden of showing that (his) right to issuance of the writ is ‘clear and indisputable.” ’ Banker's Life & Cas. Co. v. Holland, 346 U.S., at 384, 74 S.Ct., at 148, quoting United States ex rel. Bernardin v. Duell, 172 U.S. 576, 582, 19 S.Ct. 286, 287, 43 L.Ed. 559 (1899); Will v. United States, supra, 389 U.S., at 96, 88 S.Ct., at 274. Moreover, it is important to remember that issuance of the writ is in large part a matter of discretion with the court to which the petition is addressed. Schlagenhauf v. Holder, 379 U.S. 104, 112 n. 8, 85 S.Ct. 234, 239, 13 L.Ed.2d 152 (1964); Parr v. United States, supra, 351 U.S., at 520, 76 S.Ct., at 917. See also Technitrol, Inc. v. McManus, 405 F.2d 84 (C.A.8 1968), cert. denied, 394 U.S. 997, 89 S.Ct. 1591, 22 L.Ed.2d 775 (1969); Pacific Car & Foundry Co. v. Pence, 403 F.2d 949 (C.A.9 1968). Kerr, 426 U.S. at 403. Nor is mandamus a substitute for appeal, “even though hardship may result from delay and perhaps unnecessary trial.” Kerr 426 U.S. at 404 (citing Ex parte Fahey, 332 U.S. at 259-260; Bankers Life & Casualty Co. v. Holland, 346 U.S. at 382-383; United States Alkali Export Ass'n v. United States, 325 U.S. 196, 65 S.Ct. 1120, 89 L.Ed. 1554,; Evaporated Milk Ass'n, 319 U.S. at 31). “The writ is appropriately issued, however, when there is ‘usurpation of judicial power’ or a clear abuse of discretion.” Bankers Life & Casualty Co. v. Holland, 346 U.S. at 383.[1] Here, Merck has not shown that Judge Fallon committed any usurpation of judicial power or a clear abuse of discretion. Given the situation he confronted, Judge Fallon used his judicial powers diligently and within sound discretion. The record does not reflect that Merck was prejudiced to any substantial degree by the manner in which Judge Fallon conducted the in camera examination. It is true that he expected to receive the documents already divided into categories and do random samples of each category. However, he was surprised to find that Merck had only categorized the documents electronically. Thus, it appears that if anyone is responsible for the categorical plan not working it was Merck, which should have foreseen that the judge could not efficiently jump around from box to box constructing the categories before he could even begin the in camera review. There is nothing in the record that indicates that the parties and Judge Fallon had agreed that under the categorical plan Merck would have had the right to challenge the judge's determination of whether a particular category should have been designated as privileged rather than non-privileged. As the PSC points out, Merck may well have benefitted from Judge Fallon's careful document-by-document examination, for sampling necessarily would not detect all of the privileged documents in each category. *9 Significantly, Merck does not ask this court to reverse Judge Fallon's individual rulings in any respect. Instead, it asks us to examine two of its exhibits and infer from them that almost all of the district court's 31,000 some odd rulings are presumptively in error, entitling it to an order that the district court defend its classification after allowing Merck to supplement its privilege claim with additional data. In exhibit 14, a table prepared by Merck sets forth examples of allegedly inconsistent privilege rulings for 200 documents, in which the district court allegedly ruled the document privileged in one instance but not privileged in another. In exhibit 22, Merck submits approximately 50 documents (filed under seal) and summary tables, as to which Merck contends the district court's privilege rulings were erroneous. Merck's showing in respect to both exhibits is not sufficient to warrant mandamus. Exhibit 14 is nothing more than a tabular representation of the instances in which Merck alleges inconsistent rulings. Merck does not supply us with the actual documents, and the table itself admits that “some of the privilege log entries for the duplicates vary slightly in their wording.” Without recourse to the actual documents Judge Fallon examined, I am unable to determine that his rulings were inconsistent as alleged; nor can I verify that the “slight” differences in the privilege log descriptions could not have affected his determinations. Exhibit 22 is equally insufficient. I am unable to determine that Judge Fallon erred in his individual rulings to the extent that Merck contends. Merck has submitted under seal documents it claims were wrongly classified, but I find I encounter difficulties in determining whether a particular document should have been privileged. The documents are unwieldy to manage and interpret; for some, the Bates numbers are not visible, making it difficult to compare them to the privilege log. Based on this experience alone, it is hard to see how Judge Fallon's judgments were clearly in error on all of these “sample” documents. Although I do not doubt Merck's sincerity or its conviction in believing that Judge Fallon committed the errors alleged, we as appellate judges are simply not given enough information to find clear error of fact or law with respect to his privilege determinations on all of the documents in the “representative sample”. Moreover, the majority apparently agrees that Merck's showing is insufficient in this regard, as it nowhere contends that it examined the documents and determined that the district court erred in determining them to be unprivileged. As for whether Judge Fallon was able to competently examine and rule upon such a large number of documents within a two week period, I am not convinced by this record that an able, hard-working district judge could not be equal to this task. With due respect for the arguments of learned counsel, it is not self-evident to me that Judge Fallon's in camera document examination must be presumed to have been sub-standard. Merck has presented no disinterested expert opinion to this effect. Having observed Judge Fallon's work as a trial and appellate lawyer and as a federal district judge over some 30 years, I am strongly inclined to believe that he has the ability, determination and self-discipline to perform excellent judicial work under almost any set of circumstances, including the situation presented here. As everyone knows, and Merck concedes, classifying documents containing communications of employees and attorneys of large corporations for entitlement to the attorney-client privilege frequently calls for close, tough judgment calls. Laying aside the instances in which Merck and Judge Fallon may reasonably disagree upon the proper classification of the documents, the incidence of error in his rulings does not appear to be excessive. Again, as Merck and Judge Fallon both acknowledge, when dealing with such a large volume of documents, human error is bound to occur. Just as Merck honestly and in good faith initially mis-classified over 220,000 documents out of millions, Judge Fallon arguably may have made mistakes in examining some 31,846 documents within an acceptable margin of error. Slight imperfections of this kind are unavoidable in the examination and classification of a mammoth corporate privilege claim and clearly do not warrant an appellate court's mandamus. *10 The majority's re-examination order has no warrant in the record or basis in law, as evidenced by the way in which the 2,000 document figure was arrived at. In response to a judge's question in our telephonic conference call-“exactly what do you want the court to do”-Merck asserted that it wanted us to order the district court to re-examine 2,000 of the documents classified as unprivileged. On further inquiry, Merck first indicated that these were “higher priority” privileged documents, but when asked why it had not immediately called the 2,000 priority items to the district court's attention, Merck had to admit that the 2,000 figure was more or less a guess at what it needed reexamined to make sure there was no truly prejudicial disclosure of a privileged document. Further, Merck said it wanted the district court to re-examine an unspecified group of 2,000 documents in the hopes of finding significant numbers misclassified, further acknowledging that if sufficient numbers were found to be wrongly classified it might seek review beyond those 2,000. The majority commendably gives Merck neither what it originally sought nor all of its fallback, but it still grants Merck substantially scaled down version of what it proposed telephonically-district court re-examination of 2,000 documents of Merck's choosing with Merck able to furnish additional ex parte support for its privileges upon request by the district court. Ultimately, after saddling the district court with reexamination of 2,000 documents at Merck's request, the majority leaves the district judge free to make his own decision as to his privilege rulings on their merits and as to whether Merck's additional ex parte support will be considered, and thus satisfies itself that no writ has been issued at this time. Had a different course been taken, such as simply suggesting a procedure to the district court while denying mandamus or even requesting its response to our inquiry as to the feasibility of a procedure before taking action on the petition, I might have joined. But this court's ordering the district court to reexamine 2,000 documents at Merck's direction on the basis of the present record and Merck's showing in my opinion exceeds our authority. It may be that I am lacking in knowledge of the “real world” of MDL discovery on which my colleagues rely on so heavily, but in many years of experience in appellate court service I have witnessed too often the vicissitudes and folly experienced by courts at that level in attempting to closely supervise discovery and other matters within the trial courts' discretion. Finally, I agree that we have no jurisdiction to hear this case on appeal. Subject to specific exceptions, this court's jurisdiction is limited to the review of “final decisions of the district courts of the United States ... except where a direct review may be had in the Supreme Court.” 28 U.S.C. § 1291 (2000). As a general matter, discovery orders do not constitute final decisions under § 1291, and therefore, are not immediately appealable. Piratello v. Philips Electronics North America Corp., 360 F.3d 506, 508 (5th Cir.2004) (citing A-Mark Auction Galleries, Inc. v. Heritage Capital Corp., 233 F.3d 895, 897 (5th Cir.2000); Church of Scientology v. United States, 506 U.S. 9, 18 n. 11, 113 S.Ct. 447, 121 L.Ed.2d 313, (1992)). The Supreme Court has held that a party that wishes to immediately appeal a discovery order “must [first] refuse compliance, be held in contempt, and then appeal the contempt order.” Church of Scientology, 506 U.S. at 18 n. 11 (citing United States v. Ryan, 402 U.S. 530, 91 S.Ct. 1580, 29 L.Ed.2d 85, (1971)).[2] *11 In footnote 1 of its opinion, the majority correctly determines that we do not have appellate jurisdiction of Merck's appeal from the district court's interlocutory discovery and privilege ruling order (citing Texaco Inc. v. Louisiana Land and Exploration Co., 995 F.2d 43, 48 (5th Cir.1993)) That Texaco case was followed in Goodman v. Harris County, 443 F.3d 464, 467 (5th Cir. (Tex.) Mar 21, 2006), citing it for the proposition that discovery orders are generally not appealable, unless they fit within a particular exception to the collateral order rule. We are bound by the precedent of this circuit. Thus, since the court lacks jurisdiction, this panel should dismiss the appeal and the motion for emergency stay pending appeal for lack of jurisdiction. Footnotes [1] Merck urges that the privilege rulings are appealable collateral orders under Cohen v. Beneficial Industrial Loan Corp., 337 U.S. 541, 69 S.Ct. 1221, 93 L.Ed. 1528 (1949). Our current jurisprudence is contrary, and we pause only to note its questionable footing-relying as it does on the general principle that discovery rulings are not Cohen orders because they are reviewable on appeal of a final judgement in the case, failing to acknowledge the conclusive character of orders rejecting a claim of attorney-client privilege. Texaco Inc. v. Louisiana Land And Exploration Co., 995 F.2d 43, 44 (5th Cir.1993) held that an order requiring production of documents claimed to be privileged was not appealable. Texaco relied exclusively on Honig v. E.I. Du Pont De Nemours & Co., 404 F.2d 410 (1968) (per curiam). However, Honig was “... an appeal by a witness not a party to the principal lawsuit, from an order of the trial court requiring him to submit to further examination by deposition....” The court held that “... this appeal must be dismissed under the general rule that a discovery order incident to a pending action is not subject to appeal. See 6 Moore, Federal Practice, Sec. 54.07, and cases cited. This is not a final order, and we find no facts in the record that would warrant our invoking the exception to the rule as outlined in Overby v. United States Fidelity and Guaranty Co., 224 F.2d 158. (5th Cir.1955).”). The dissent here also relies on a case which fails to distinguish rulings on the attorney-client privilege, failing to address an issue that has divided the circuits. Compare In re Ford Motor Co., 110 F.3d 954, 957-64 (3d Cir.1997) (collateral-order review available to challenge order denying claim of attorney-client privilege); United States v. Philip Morris, Inc., 314 F.3d 612, 617-21 (D.C.Cir.2003) (discovery order denying claim of attorney-client privilege immediately appealable as a collateral order, particularly since civil contempt is not amenable to immediate appeal in the Circuit); Agster v. Maricopa County, 422 F.3d 836, 838 (9th Cir.2005) (recognizing but declining to broach the open question of the appealability of adverse privilege determinations in general under the collateral order doctrine, while finding jurisdiction for the specific privilege at issue); with FDIC v. Ogden Corp., 202 F.3d 454, 458-460 & n. 2 (1st Cir.2000) (substantial privilege claim that cannot effectively be tested by privilege-holder through contemptuous refusal ordinarily will qualify for immediate review if claim otherwise would be lost); Chase Manhattan Bank v. Turner & Newall, PLC, 964 F.2d 159, 162-63 (2d Cir.1992) (order compelling production of privileged documents without judicial ruling on merits not appealable but reviewable under writ of mandamus); Ross v. City of Memphis, 423 F.3d 596, 599 (6th Cir.2005) (noting that in the case of attorney-client privilege an order compelling testimony is ordinarily not appealable, however, an exception exists for those situations where a party claiming a privilege is powerless to avert the mischief of the order) (internal quotation omitted); Reise v. Board of Regents of the Univ. of Wis. Sys., 957 F.2d 293, 295 (7th Cir.1992) (“[E]ven orders to produce information over strong objections based on privilege are not appealable, despite the claim that once the cat is out of the bag the privilege is gone.”); Boughton v. Cotter Corp., 10 F.3d 746, 750 (10th Cir.1993) (adverse determination concerning attorney-client privilege not appealable as collateral order); Quantum Corp. v. Plus Dev. Corp., 940 F.2d 642, 644 (Fed.Cir.1991). [2] The privilege log identified for every withheld document: the bates number; the date of the communication, the names of the author(s) and/or correspondent(s); the names of the recipient(s) and/or addressee(s); a description of the document, including the subject matter upon which legal advice was sought or given; the nature of the privilege asserted; and the attachments related to that document. [3] 220 F.3d 568, 572 (7th Cir.2000). [4] As there is no turning back from the disclosure of the documents, the need for a careful determination at this juncture is self-evident, else rendering the decision unreviewable by any other means to effectively protect the claimed privilege of thousands of documents. See, e.g., In re Burlington N., Inv., 822 F.2d 518, 522-23 (5th Cir.1987) (stating that erroneous disclosure of privileged documents could be irreparable); see also In re Occidental Petroleum Corp., 217 F.3d 293, 295 n. 9 (5th Cir.2000) (acknowledging that an erroneous privilege determination during discovery may prove the “necessity” for review) (citing In re Spalding Sports Worldwide, Inc., 203 F.3d 800, 804 (Fed.Cir.2000) (observing that “... maintenance of the attorney-client privilege up to its proper limits has substantial importance to the administration of justice, and ... an appeal after disclosure of the privileged communication is an inadequate remedy”) (internal quotation omitted)); In re General Motors Corp., 153 F.3d 714, 715 (8th Cir.1998) (noting that “... the district court's order would otherwise destroy the confidentiality of the communications at issue.”)). [5] By random sampling, we mean adhering to a statistically sound protocol for sampling documents. See generally Am. Nat'l Bank & Trust Co. v. Equitable Life Assur. Soc'y, 406 F.3d 867, 879 (7th Cir.2005). The parties must provide expert assistance to the district court in constructing any protocol. In order to maximize the utility of this exercise, the process should return the parties to the categorical approach, Merck having the responsibility to generate a set of categories simultaneously representative of both the groups of 2,000 and 30,000 documents (this presupposes that the categories will be sufficiently large to support a random sampling). [6] See United States v. Zolin, 491 U.S. 554, 109 S.Ct. 2619, 2629, 105 L.Ed.2d 469 (1989) (“ ... [T]his court has approved the practice of requiring parties who seek to avoid disclosure of documents to make the documents available to in camera inspection”); In re Grand Jury Investigation, 974 F.2d 1068, 1071 (9th Cir.1992). [1] The circuit precedents upon which the majority relies are inapposite. In In re Occidental Petroleum Corp., 217 F.3d 293, 295 (5th Cir.2000) this court denied the petition for mandamus, “[f]inding no clear error, because the plaintiff class [as a shareholder] is not subject to Occidental's attorney-corporate client privilege under Garner v. Wolfinbarger, 430 F.2d 1093, 1103-4 (5th Cir.1970).” In re Burlington N., Inv., 822 F.2d 518, 522-23 (5th Cir.1987) is inapposite because it presented truly “exceptional circumstances.” “[U]nlike the typical discovery order, the district court's order was not a mere discretionary one but rather turns on legal questions appropriate for appellate review.... Given the nature of the analysis required in an antitrust case in which a Noerr-Pennington defense is asserted and a question of sham is raised, communication between the attorney directing the petitioning activity and his client will be highly relevant. Under the district court's ruling, such communications could be discovered upon a showing that the petitioning was part of a larger conspiracy even though the petition itself might be protected from antitrust scrutiny. Such an important and potentially far-reaching decision, which we hold below to be erroneous, is an appropriate one for our immediate review.” Id. at [ ] [2] The rule is different, however, where the person asserting privilege is a non-party. See Church of Scientology, 506 U.S. at 18 n. 11. Further, the Supreme Court has carved out an exception for pre-contempt appeals by the President of the United States to avoid unnecessary constitutional confrontations between two coordinate branches of government. See United States v. Nixon, 418 U.S. 683, 94 S.Ct. 3090, 41 L.Ed.2d 1039 (1974).