Denmark v. Liberty Life Assurance Company of Boston

United States Court of Appeals For the First Circuit No. 05-2877 DIANE DENMARK, Plaintiff, Appellant, v. LIBERTY LIFE ASSURANCE COMPANY OF BOSTON, Defendant, Appellee. APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF MASSACHUSETTS [Hon. Douglas P. Woodlock, U.S. District Judge] Before Lipez, Selya and Howard, Circuit Judges. Jonathan M. Feigenbaum, with whom Phillips & Angley was on brief, for appellant. Jay E. Sushelsky and Melvin R. Radowitz on brief for American Association of Retired Persons, amicus curiae. Mala M. Rafik and Rosenfeld & Rafik, P.C. on brief for Massachusetts Employment Lawyers Association, amicus curiae. Richard Johnston on brief for Health Administration Responsibility Project, amicus curiae. Andrew C. Pickett, with whom Matthew D. Freeman, Ashley B. Abel, and Jackson Lewis LLP were on brief, for appellee. Lisa Tate, Teresa L. Jakubowski, Mark J. Crandley, and Barnes & Thornburg, LLP on brief for American Council of Life Insurers, amicus curiae. May 6, 2009 SELYA, Circuit Judge. This appeal has generated thorny questions involving the appropriate standard of judicial review under the Employee Retirement Income Security Act (ERISA), 29 U.S.C. §§ 1001-1461. It is now before us for a second time. Our initial encounter produced a proliferation of views: three separate opinions from the three panelists, each of which grappled with the methodological problem facing a reviewing court in regard to an ERISA benefit-denial decision made by a plan administrator operating as both adjudicator and payer of such claims. See Denmark v. Liberty Life Assur. Co., 481 F.3d 16 (1st Cir. 2007) (Lipez, J.); id. at 39 (Selya, J., concurring); id. at 41 (Howard, J., dissenting). For ease in exposition, we refer to the three constituent opinions comprising that splintered decision, collectively, as "Denmark II." Dissatisfied with the outcome, the plaintiff sought rehearing and rehearing en banc. See Fed. R. App. P. 35; 1st Cir. R. 35, 40. The en banc court withheld action on the petition until the Supreme Court had decided Metropolitan Life Insurance Co. v. Glenn, 128 S. Ct. 2343 (2008). Believing that Glenn had shed new light on the standard of review, the panel withdrew its earlier decision and requested supplemental briefing. See Denmark v. Liberty Life Assur. Co., 530 F.3d 1020 (1st Cir. 2008) (per curiam). By separate order, the petition for rehearing en banc was denied as moot. -2- The supplemental submissions, together with a welter of helpful amicus briefs, led to a new round of oral argument. We took the matter under advisement and now reaffirm our existing abuse of discretion standard of review, albeit with certain refinements. We nonetheless recognize that the ultimate resolution of the case may be informed, under Glenn, both by the aforementioned refinements and by the obtaining of further information. Consequently, we vacate the judgment and remand to the district court so that it may obtain that information and reevaluate the case with the guidance supplied by Glenn and by this opinion. I. BACKGROUND We presume the reader's familiarity with the facts of the case as set forth in Denmark II. We rehearse here only those events necessary to put this appeal, in its present posture, into a workable perspective. In 1996, a primary care physician diagnosed plaintiff- appellant Diane Denmark as suffering from fibromyalgia. The plaintiff, who was a group leader employed by GenRad, Inc., nonetheless continued to work. At the times relevant hereto, she was covered under two interlocking, ERISA-regulated disability insurance plans: GenRad's short-term disability plan (the STD Plan) -3- and its long-term disability plan (the LTD Plan).1 Defendant- appellee Liberty Life Assurance Company (Liberty) administered both plans, albeit under different arrangements. The employer self-funded the STD Plan. Under it, Liberty provided an initial claims review and benefits determination. Its decisions were appealable to the employer, which paid approved claims from its own exchequer. In contrast, Liberty underwrote the LTD Plan. Pursuant to its terms, Liberty reviewed all claims, made the initial benefits determinations, adjudicated any appeals, and paid approved claims from its own coffers. The plaintiff stopped working on October 3, 2001, and applied for STD benefits. The STD Plan defines "disabled" to include a person who is "unable to perform all of the material and substantial duties of [her] occupation . . . because of an Injury or Sickness." In an effort to satisfy this definition, the plaintiff supported her claim with reports from three doctors: her primary care physician, a cardiologist, and a rheumatologist. After reviewing the tendered medical records and a job description, Debra Kaye, a nurse employed by Liberty as a case manager, requested that Dr. Clay Miller conduct a peer review. Based on Dr. Miller's assessment, Liberty denied the claim. 1 In late 2001, Teradyne, Inc. acquired GenRad, but the plaintiff's right to coverage remained the same. For simplicity's sake, we refer throughout to GenRad. -4- The plaintiff appealed this decision to her employer. The appeal papers included a response from her primary care physician disputing Dr. Miller's conclusions. The employer asked Dr. Peter Schur to perform an independent medical examination (IME). When Dr. Schur found the plaintiff disabled, the employer agreed to pay her STD benefits. In June of 2002, the plaintiff filed for long-term benefits. An applicant qualifies as disabled under the LTD Plan if, for the first two years, "as a result of Injury o[r] Sickness, [she] is unable to perform the Material and Substantial Duties of [her] Own Occupation" and thereafter "is unable to perform, with reasonable continuity, the Material and Substantial Duties of Any Occupation." Nurse Kaye reviewed the file, which contained medical support for a finding that the plaintiff's symptomatology had worsened as well as a completed activities questionnaire in which she claimed to have severe restrictions on her ability to sit, stand, walk, drive, and concentrate. In her second review, Nurse Kaye discounted the IME report, suggested that the plaintiff's condition was not as grave as the completed questionnaire implied, and concluded that the plaintiff did not qualify for LTD benefits. Thus, Liberty denied the claim. The plaintiff requested further review. Liberty responded by, among other things, determining that her job involved -5- light to sedentary work and hiring a private investigator to surveil the plaintiff's activities. The sleuth furnished written reports and photographs showing that the plaintiff was "very active." With this ammunition in hand, Liberty submitted the entire file to Network Medical Review (NMR), a referral service furnishing physicians to evaluate the functional abilities of claimants. NMR forwarded the assignment to one of its correspondents, Dr. John Bomalaski, who concluded that the plaintiff was capable of working full-time in her (primarily sedentary) position. On December 10, 2002, Liberty reaffirmed its earlier denial of LTD benefits. Nearly fourteen months later, an administrative law judge ruled the plaintiff entitled to social security disability benefits, see 42 U.S.C. § 405, retroactive to her last day of actual work. The judge premised this decision on a subsidiary finding that the plaintiff was disabled within the meaning of the Social Security Act. See id. § 423(d); see also 20 C.F.R. § 416.920. Although the definition of disability under the Act differed from the definition of disability under the LTD Plan, the plaintiff transmitted this ruling, along with a further report from her rheumatologist, to Liberty; based on these documents, she sought reconsideration of the refusal to pay LTD benefits. Liberty stood firm. -6- II. TRAVEL OF THE CASE On September 17, 2004, the plaintiff sued Liberty in a Massachusetts state court. Liberty removed the action to the federal district court. See 28 U.S.C. § 1441; see also id. § 1331. The case proceeded on the plaintiff's claim under 29 U.S.C. § 1132(a)(1)(B). The district court permitted the plaintiff to conduct limited discovery anent Liberty's relationship with NMR and its correspondent physicians as part of an effort to show that Liberty's actions were influenced by a conflict of interest. Liberty acknowledged that it had paid upwards of $2,000,000 to NMR physicians between 2001 and 2003, and identified 1,204 files that it had referred to NMR during that interval. But Liberty refused, on burdensomeness grounds, to answer interrogatories regarding the proportion of those files in which claims ultimately had been allowed. As a sanction for this recalcitrance, the court drew an inference that NMR had found against the claimants in all cases and, thus, applied heightened scrutiny to Dr. Bomalaski's opinion. Denmark v. Liberty Life Assur. Co. (Denmark I), Civ. No. 04-12261, 2005 WL 3008684, at *11 (D. Mass. Nov. 10, 2005). In due season, the parties cross-moved for summary judgment. Noting that the plan documents delegated discretionary authority to Liberty, qua plan administrator, the court reviewed the benefit-denial decision under this circuit's historic abuse of -7- discretion standard. Id. at *9. Although the court acknowledged the potential conflict of interest posed by Liberty's dual role in making benefits determinations and paying claims,2 it found no significantly probative evidence that the conflict had in fact influenced Liberty's decisionmaking. Id. at *18. In discussing that issue, the court opined that a bare structural conflict, in and of itself, did not warrant the application of a less deferential standard of review. Id. at *9. The court proceeded to find the denial of LTD benefits supported by substantial evidence and, thus, within the plan administrator's discretion. Id. at *26. Accordingly, it granted Liberty's summary judgment motion and denied the plaintiff's. See id. On appeal, the plaintiff pursued two lines of attack. First, she contended that the district court had employed an incorrect standard of review. Second, she contended that, whatever the standard of review, the denial of LTD benefits was insupportable. We have recounted the rest of the tale above: the panel, by a divided vote, affirmed the district court's ruling; the plaintiff petitioned for rehearing; the Supreme Court decided Glenn; and the litigation then entered its current phase. 2 We call such instances structural conflicts, in contradistinction to actual conflicts (i.e., instances in which the fiduciary's decision was in fact motivated by a conflicting interest). -8- III. ANALYSIS The focal point of this appeal has become the standard of judicial review. For that reason, we think it useful to rehearse how the case law in that area has evolved. Among its panoply of remedial devices for plan participants, ERISA provides for suits to enforce rights conferred under the terms of an ERISA-regulated plan. See 29 U.S.C. § 1132(a)(1)(B). Suits for the recovery of benefits come within the ambit of this provision. Congress did not elucidate a standard of judicial review applicable to such actions. The Supreme Court filled this void when it decided Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101 (1989). The Firestone Court noted that ERISA "abounds with the language and terminology of trust law" and that Congress anticipated the development of a "federal common law of rights and obligations under ERISA-regulated plans." Id. at 110. Invoking trust principles, the Court held that when an ERISA-regulated plan vests discretion in the plan administrator, the latter's resolution of benefits claims must be reviewed deferentially. Id. at 111. Absent such a delegation of discretionary authority, a plan administrator's decisions are to be reviewed de novo. Id. at 111-12. In a brief aside, the Court observed that "if a benefit plan gives discretion to an administrator or fiduciary who is operating under a conflict of interest, that conflict must be -9- weighed as a 'facto[r] in determining whether there is an abuse of discretion.'" Id. at 115 (dictum; quoting Restatement (Second) of Trusts § 187 cmt. d (1959)). For the next eighteen years, courts struggled both with this dictum and with how to handle structural conflicts of interest in ERISA cases. A number of different approaches emerged. This court clung to the classic abuse of discretion model, taking account of the impact, if any, of a conflict in evaluating whether a denial of benefits was arbitrary and capricious (and thus, an abuse of discretion). See, e.g., Leahy v. Raytheon Co., 315 F.3d 11, 15-16 (1st Cir. 2002); Doe v. Travelers Ins. Co., 167 F.3d 53, 57-58 (1st Cir. 1999). Other circuits, however, adopted divergent approaches. See Kathryn J. Kennedy, Judicial Standard of Review in ERISA Benefit Claim Cases, 50 Am. U. L. Rev. 1083, 1135-72 (2001) (collecting cases). This compendium included a "presumptive neutrality" approach, under which abuse of discretion review obtains except in cases of actual conflict (that is, cases in which the plan administrator's decision is shown to be conflict-driven). See, e.g., Kobs v. United Wis. Ins. Co., 400 F.3d 1036, 1039 (7th Cir. 2005); Pulvers v. First Unum Life Ins. Co., 210 F.3d 89, 92 (2d Cir. 2000); Woo v. Deluxe Corp., 144 F.3d 1157, 1160-61 (8th Cir. 1998). It also included a "combination of factors" approach under which abuse of discretion review treats both actual and potential -10- conflicts of interest as relevant factors. See, e.g., Abatie v. Alta Health & Life Ins. Co., 458 F.3d 955, 965-69 (9th Cir. 2006); Calvert v. Firstar Fin. Inc., 409 F.3d 286, 293 (6th Cir. 2005). Several courts favored a "sliding-scale" approach. See, e.g., Pinto v. Reliance Standard Life Ins. Co., 214 F.3d 377, 391-92 (3d Cir. 2000); Vega v. Nat'l Life Ins. Servs., Inc., 188 F.3d 287, 296 (5th Cir. 1999) (en banc); Chambers v. Family Health Plan Corp., 100 F.3d 818, 825-26 (10th Cir. 1996); Doe v. Group Hosp'n & Med. Servs., 3 F.3d 80, 87 (4th Cir. 1993). One court preferred a six-step burden- shifting approach. See Williams v. BellSouth Telecomms., Inc., 373 F.3d 1132, 1138 (11th Cir. 2004). Because our home-grown standard was central to the decisions in both Denmark I and Denmark II, we explore its parameters. In Doyle v. Paul Revere Life Insurance Co., 144 F.3d 181 (1st Cir. 1998), we acknowledged that the Firestone dictum could be read to imply a heightening of the standard of review for structural conflict cases. Id. at 184. We noted, however, that market forces were at work: employers are unlikely to contract with insurers who acquire reputations for miserliness. Id. Thus, it seemed prudent to adhere to the baseline abuse of discretion standard in cases involving structural conflicts, but to give that standard "more bite"; that is, a "special emphasis on reasonableness." Id. The bottom-line inquiry should be "whether [the plan administrator] had substantial evidentiary grounds for a reasonable decision in its -11- favor." Id. This approach left the claimant free to show an actual conflict — and if she succeeded in doing so, that showing would influence the decisional calculus. In Doe, 167 F.3d at 57-58, we supplied a gloss on Doyle, explaining that reasonableness "is the basic touchstone" in all benefit-denial cases. We again rejected a special standard of review for structural conflict cases, observing that "gradations in phrasing are as likely to complicate as to refine the standard." Id. In any event, the requirement of reasonableness is flexible; thus, that requirement may have "substantial bite" when a court is faced with a specific decision on a specific set of facts. Id. (explaining that reasonableness review necessarily takes cognizance of conflicts). In Pari-Fasano v. ITT Hartford Life & Accident Insurance Co., 230 F.3d 415 (1st Cir. 2000), we stressed two points. The first dealt with nomenclature; we made pellucid that the terms "abuse of discretion," "arbitrary and capricious," and "reasonableness" were functionally equivalent in the ERISA context. Id. at 419. None of those terms heralded a heightened standard of review for structural conflict cases. Id. Our second point remarked the obvious: "the possible existence of a conflict of interest would necessarily affect the court's determination of what was reasonable conduct by the insurer under the circumstances." Id. When Pari-Fasano speaks of the potential for conflict, we understand -12- that usage as a reference to the existence of a structural conflict, and not the possibility of finding an actual conflict or "improper motivation." Id.3 Following this trilogy of cases, we consistently have reviewed the resolution of benefits claims by structurally conflicted plan administrators for abuse of discretion, taking into account both the potential for conflict and the mitigating effect of market forces.4 See, e.g., Buffonge v. Prudential Ins. Co., 426 F.3d 20, 28 & n.11 (1st Cir. 2005); Wright v. R.R. Donnelley & Sons Co. Group Benefits Plan, 402 F.3d 67, 74 (1st Cir. 2005); Glista v. Unum Life Ins. Co., 378 F.3d 113, 125-26 (1st Cir. 2004); Lopes v. Metro. Life Ins. Co., 332 F.3d 1, 4-5 (1st Cir. 2003); Leahy, 315 F.3d at 16; Dandurand v. Unum Life Ins. Co., 284 F.3d 331, 335-36 (1st Cir. 2002). This brings us to Glenn. There, the Supreme Court reviewed a denial of benefits by an administrator that both passed judgment upon and paid claims under an ERISA-regulated plan. The 3 Insofar as our later cases read this language as precluding consideration of a purely structural conflict in assessing the existence vel non of an abuse of discretion, that interpretation is inconsistent with Glenn, 128 S. Ct. at 2351. 4 In Leahy, we suggested that when a plan administrator's determination is actually motivated by a conflict of interest, "courts may cede a diminished degree of deference — or no deference at all — to the administrator's determinations." 315 F.3d at 16. When faced with such a case, we determined that the situation warranted de novo review. See Janiero v. Urological Surgery Prof'l Ass'n, 457 F.3d 130, 139-42 (1st Cir. 2006). We need not speculate here as to whether the holding in Janiero survives Glenn. -13- denial had been upheld by the district court but set aside by the Sixth Circuit under its "combination of factors" standard of review. Picking up on the Firestone dictum, the Glenn Court clarified what sort of relationships might suffice to create a conflict of interest. It concluded that courts should take cognizance of structural conflicts in ERISA cases; that is, that a conflict exists whenever a plan administrator, whether an employer or an insurer, is in the position of both adjudicating claims and paying awarded benefits. Glenn, 128 S. Ct. at 2348-50. In reaching that conclusion, the Court rejected the market forces rationale, explaining that "ERISA imposes higher-than-marketplace quality standards on insurers." Id. at 2350. The Court left open the possibility that market forces might inform the significance of a structural conflict in a given case. See id. The Court then turned to the question of how best to weigh structural conflicts. In charting this course, it held fast to the standard of review previously announced in Firestone: abuse of discretion. Id. (analogizing to trust law, which asks merely whether a conflicted trustee has abused his discretion either substantively or procedurally). The Court rejected burden-shifting rules as a mechanism for ensuring proper judicial review of decisions made by structurally conflicted plan administrators. Id. at 2351. -14- On a more affinitive note, the Court commented approvingly that "when judges review the lawfulness of benefits denials, they will often take account of several different considerations of which a conflict of interest is one." Id. It likened this multi-factor approach to that used in the administrative law context.5 Id. The Court added that judges should weigh a conflict as they would weigh any other pertinent factor; that is, when the relevant considerations are in equipoise, any one factor, including a structural conflict, may act as a tiebreaker. Id. In this regard, the Court counselled judges to take account of both "the degree of closeness" and "the tiebreaking factor's inherent or case-specific importance." Id. The Court acknowledged the resemblance between its approach and the Sixth Circuit's "combination of factors" approach; those approaches give a structural conflict some weight but, in the absence of aggravating circumstances (say, evidence of arbitrariness or of actual bias), do not treat it as a dispositive influence. Id. at 2351-52. 5 Seizing on this comparison and the Court's use of the term "lawfulness," the plaintiff suggests that the Court created a new, less deferential standard of review. But the Court's articulation of how trust law informs the issue reveals the utter implausibility of this suggestion. See Glenn, 128 S. Ct. at 2350 ("Trust law continues to apply a deferential standard of review to the discretionary decisionmaking of a conflicted trustee, while at the same time requiring the reviewing judge to take account of the conflict when determining whether the trustee, substantively or procedurally, has abused his discretion.") (emphasis supplied)). -15- The Court also described what kind of evidence might impact the relative weight of an identified conflict: The conflict of interest at issue here, for example, should prove more important (perhaps of great importance) where circumstances suggest a higher likelihood that it affected the benefits decision, including, but not limited to, cases where an insurance company administrator has a history of biased claims administration. It should prove less important (perhaps to the vanishing point) where the administrator has taken active steps to reduce potential bias and to promote accuracy, for example, by walling off claims administrators from those interested in firm finances, or by imposing management checks that penalize inaccurate decisionmaking irrespective of whom the inaccuracy benefits. Id. at 2351 (citations and internal quotations marks omitted). To complete the picture, the Court applied its newly refined standard to the case before it. In so doing, the Court assessed a litany of relevant factors, including the plan administrator's structural conflict, its inconsistent positions concerning a social security determination, its unexplained emphasis on medical opinions favoring a denial of benefits, and its offhand discounting of contrary medical opinions. Id. at 2352. The Court concluded that "these serious concerns," together with the closeness of the case and the presence of a structural conflict, supported the decision to set aside the plan administrator's discretionary judgment. Id. The case at bar falls squarely within Glenn's precedential orbit. Here, the LTD Plan contains a sufficient delegation of -16- discretionary authority to trigger deferential review. See Denmark II, 481 F.3d at 29; id. at 40 (Selya, J., concurring); id. at 41 (Howard, J., dissenting). That brings into play Glenn's baseline principle, consistent with this circuit's prior precedent, that judicial review of such a benefit-denial decision is for abuse of discretion. See Glenn, 128 S. Ct. at 2350; see also Doe, 187 F.3d at 56-57; Doyle, 144 F.3d at 184. In other words, where the plan documents delegate discretionary authority to the plan administrator (whether or not structurally conflicted), courts should review benefit-denial decisions for abuse of discretion, considering any conflict as one of a myriad of relevant factors. See Glenn, 128 S. Ct. at 2351. At this point, a red flag appears. Although the standard of review articulated in our earlier cases comports generally with Glenn, two aspects of our original approach require refinement. First, the market forces rationale no longer allows a reviewing court to disregard a structural conflict without further analysis. See Glenn, 128 S. Ct. at 2349-50. That aspect of the Glenn decision requires that structural conflicts be accorded weight — albeit not necessarily dispositive weight — in the standard-of-review equation. With that in mind, courts are duty-bound to inquire into what steps a plan administrator has taken to insulate the decisionmaking process against the potentially pernicious effects of structural conflicts. -17- Second, Glenn makes explicit what was implicit in our earlier decisions: in cases in which a conflict has in fact infected a benefit-denial decision, such a circumstance may justify a conclusion that the denial was itself arbitrary and capricious (and, thus, an abuse of discretion). See id. at 2351; McCauley v. First Unum Life Ins. Co., 551 F.3d 126, 138 (2d Cir. 2008). To sum up, our preexisting standard of review is largely but not entirely harmonious with Glenn. While the refinements are modest, this case is hair's-breadth close. Given that precarious balance, even a slight adjustment in the mix of factors or in the weight of a single factor may make a decisive difference. Hence, we think it incumbent upon us to remand the case and permit the district court, in the first instance, to reconsider its decision in light of Glenn.6 Remand will allow full consideration of how heavily this conflict should weigh in the balance. That is highly desirable because, in performing a multi-factor analysis, "any one factor will act as a tiebreaker when the other factors are closely balanced." Glenn, 128 S. Ct. at 2351. We leave this reweighing to the district court, and intimate no view as to the outcome. Notwithstanding our decision to remand, our journey is not yet at an end. The supplemental briefing touched upon discovery issues, see, e.g., Appellee's Br. on Reh'g at 57-58, and at oral 6 In its original decision, the district court mentioned the structural conflict, but it considered it in only a glancing way. See Denmark I, 2005 WL 3008684, at *9. -18- argument in this court the parties vigorously debated the permissible scope of discovery, post-Glenn, in ERISA cases. Denmark's counsel argued that she should be allowed "to engage in normal discovery" (which he defined by reference to the discovery that would be permitted in a personal injury suit arising out of a traffic accident) and that denying such unfettered discovery would be unfair and contrary to ERISA as seen through the prism of Glenn. Counsel for Liberty took a markedly less ambitious view of discovery in ERISA cases. Given these disparate appraisals, we have a responsibility to offer guidance to the parties and the district court. That guidance entails a brief discussion about the scope of discovery in ERISA cases. ERISA benefit-denial cases typically are adjudicated on the record compiled before the plan administrator. Because full- blown discovery would reconfigure that record and distort judicial review, courts have permitted only modest, specifically targeted discovery in such cases. See Liston v. Unum Corp. Officer Sev. Plan, 330 F.3d 19, 23 (1st Cir. 2003) (noting that "some very good reason is needed to overcome the strong presumption that the record on review is limited to the record before the administrator"). In some cases, a good reason has been found to exist when a party makes a colorable claim of bias. See id. Targeted discovery addressed to such an issue may shed new light on the -19- motivation behind the plan administrator's decision without expanding the panoply of materials on which that decision was based. The majority opinion in Glenn fairly can be read as contemplating some discovery on the issue of whether a structural conflict has morphed into an actual conflict. See, e.g., Glenn, 128 S. Ct. at 2351. That is consistent with the Liston paradigm. But any such discovery must be allowed sparingly and, if allowed at all, must be narrowly tailored so as to leave the substantive record essentially undisturbed. In future cases, plan administrators, aware of Glenn, can be expected as a matter of course to document the procedures used to prevent or mitigate the effect of structural conflicts. That information will be included in the administrative record and, thus, will be available to a reviewing court. Conflict-oriented discovery will be needed only to the extent that there are gaps in the administrative record. If, say, the plan administrator has failed to detail its procedures,7 discovery may be appropriate, in the district court's discretion. Otherwise, discovery normally will be limited to the clarification of ambiguities or to ensuring that the documented procedures have been followed in a particular instance. The case at hand falls into a special niche. Because the denial of benefits and the commencement of suit both predated Glenn, 7 These are merely exemplars; we do not pretend to canvass the entire universe of possibilities. -20- Liberty did not include in the administrative record any evidence with respect to its conflict-ameliorating procedures. Given these temporally awkward circumstances, we think that the district court, in its discretion, may wish to afford the parties a limited opportunity to flesh out the record (even if that entails further, appropriately circumscribed, discovery). IV. CONCLUSION We need go no further. For the reasons elucidated above, we vacate the judgment below and remand this case to the district court for further consideration consistent with Glenn and with this opinion. The district court is free to abrogate or modify the discovery sanction previously imposed if it sees fit to do so. One final point comes to mind. This may be an appropriate time for the parties seriously to consider settlement. The district court would be wise to explore that possibility. Vacated and remanded. No costs. - Concurring Opinion Follows - -21- LIPEZ, Circuit Judge, concurring. I agree with my colleagues that the focal point of this appeal was the standard of judicial review in our circuit in the wake of Glenn. I also agree with my colleagues that we should remand to the district court so that it can evaluate the impact of Glenn on the merits of Denmark's case. However, I am concerned that the majority's general comments about the appropriate scope of discovery post-Glenn reflect a particularly hostile attitude towards such discovery, and suggest that the issue has already been resolved in this circuit. I write separately to emphasize that it has not been resolved. Although it is true, as the majority says, that "at oral argument . . . the parties argued strenuously about the permissible scope of discovery, post-Glenn, in ERISA cases," that issue was raised sua sponte by members of the panel, not the parties. The scope of discovery post-Glenn was never part of this appeal. It was not briefed by the parties. They did not seek guidance on the issue. Instead, it is the majority that is eager to use this case to provide that guidance. It may be appropriate, in some instances, to venture beyond what is strictly required to decide a particular appeal and provide such guidance. But the resort to dicta in this case is ill- advised for two reasons. First, the issue of the permissible scope of discovery post-Glenn is complex and fact-dependent. Generalizations without context ignore that reality. Second, there -22- are cases on our doorstep from the district courts that will require us to decide these discovery issues as they should be decided -- with the benefit of district court analysis and briefing by the parties. The majority is correct that "Glenn fairly can be read as contemplating some discovery on the issue of whether a structural conflict has morphed into an actual conflict." The majority's statement that "in future cases, plan administrators . . . can be expected . . . to document the procedures used to prevent or mitigate the effect of structural conflicts" is a reasonable inference from Glenn's observation that the importance of structural conflicts is lessened where the administrator "has taken active steps to reduce potential bias and to promote accuracy." 128 S. Ct. at 2351. It is also true, as the majority notes, that in this case "the denial of benefits and the commencement of suit both predated Glenn." Therefore, on remand, "the district court, in its discretion, may wish to afford the parties a limited opportunity to flesh out the record" with "appropriately circumscribed" discovery. That general reference to "appropriately circumscribed" discovery is fair enough. The problem arises with the majority's characterizations of that appropriately circumscribed discovery. The majority says that "any such discovery [on the issue of whether a structural conflict has morphed into an actual conflict] must be allowed sparingly and, if allowed at all, must be narrowly tailored -23- so as to leave the substantive record essentially undisturbed." The majority adds that Conflict-oriented discovery will be needed only to the extent that there are gaps in the administrative record. If, say, the plan administrator has failed to detail its procedures, discovery may be appropriate, in the district court's discretion. Otherwise, discovery normally will be limited to the clarification of ambiguities or to insuring that the documented procedures have been followed in a particular instance. These propositions reflect a grudging approach to post- Glenn discovery that may not be justified. They are unnecessary for our decision in this appeal. They have been fashioned without the benefit of district court analysis or briefing by the parties. Under these circumstances, courts "are far more likely . . . to fashion defective rules, and to assert misguided propositions, which have not been fully thought through." Pierre N. Leval, Judging Under the Constitution: Dicta About Dicta, 81 N.Y.U. L. Rev. 1249, 1263 (2006). Accepted uncritically as law, such propositions can skew the decision-making process of the district courts. It is simply impossible to know in this case or in future cases the degree of discovery that may be required to establish "whether a structural conflict has morphed into an actual conflict." Such discovery might be sparing or more expansive depending upon the preliminary showing made by the plaintiff in a particular case. Decreeing in this case that such discovery must be allowed sparingly, or confined to certain categories, is an unwarranted signal that discovery into the -24- existence of an actual conflict is disfavored.8 The district court here, and our district courts generally, are fully capable of sorting through, in the first instance, the complicated discovery issues raised by Glenn, and they should not feel bound by the hostile attitude towards discovery that is improvidently reflected in dicta in the majority opinion. Those dicta are not binding on the district courts or future panels of this court. 8 This case has its own discovery history. Denmark filed a motion in the district court seeking discovery related to the financial relationship between Liberty and its "independent" review agency, NMR -- specifically, the amount of money that Liberty had paid NMR, the number of cases Liberty had referred to NMR, and how many of those claims had been granted. The district court granted the motion and ordered Liberty to produce the information Denmark requested. See Denmark v. Liberty Life Assur. Co. of Boston, 481 F.3d 16, 32 (1st Cir. 2007). Liberty provided information regarding the amount it had paid NMR between 2001 and 2003 and the number of files it had referred to them during that time period, but refused to stipulate the number of cases in which benefits had been granted on the grounds that such a stipulation would be too burdensome. Id. As a sanction for Liberty's refusal to comply with its discovery order, the district court drew the inference that NMR had not found in favor of a single claimant in all of the Liberty files it had reviewed during the relevant time period. Id. The court then stated that, in light of this inference and to account for the effects of this conflict of interest, it would review the opinion of NMR's reviewing physician with "more bite." Id. Liberty protested this sanction in its supplemental briefing. See, e.g., Appellee's Br. on Reh'g at 57-58. Aware of this history, the majority says that the district court is free on remand to abrogate or modify the discovery sanction it previously imposed. I agree. However, exactly how the district court on remand should supplement, if at all, discovery already allowed should be left to the discretion of the district court without the unwarranted signals of the majority. -25-