884 F.3d 246
5th Cir.
2018

ARIANA M., Plaintiff - Appellant v. HUMANA HEALTH PLAN OF TEXAS, INCORPORATED, Defendant - Appellee

No. 16-20174

IN THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT

March 1, 2018

Appeal from the United States District Court for the Southern District of Texas

Before STEWART, Chief Judge, and JOLLY, JONES, SMITH, DENNIS, CLEMENT, PRADO, OWEN, ELROD, SOUTHWICK, HAYNES, GRAVES, HIGGINSON, and COSTA, Circuit Judges.*

GREGG COSTA, Circuit Judge, joined by STEWART, Chief Judge, DENNIS, PRADO, SOUTHWICK, HAYNES, GRAVES, and HIGGINSON, Circuit Judges:

When an ERISA plan lawfully delegates discretionary authority to the plan administrator, a court reviewing the denial of a claim is limited to assessing whether the administrator abused that discretion. . For plans that do not have valid delegation clauses, the Supreme Court has held that “a denial of benefits challenged under § 1132(a)(1)(B) is to be reviewed under a de novo standard.” For a quarter century, we have interpreted that holding to apply only to a denial of benefits based on an interpretation of plan language. The result is a bifurcated standard of review for challenges in our circuit to the denial of ERISA benefits. Courts reviewing challenges to the legal interpretation of a plan do not, as says, give any deference to the administrator‘s view of plan language. But challenges to an administrator‘s factual determination that a beneficiary is not eligible are reviewed under the same abuse-of-discretion standard that applies when plans have delegated discretion. . When was decided, it created a circuit split with one other court of appeals that had read to set a default de novo standard for both legal and factual determinations. , overruled on other grounds by . In the time since, seven other courts of appeals have chimed in. Every one has taken thе view that the standard of review does not depend on whether the denial is deemed to be based on legal or factual grounds.

We thus have long stood alone in limiting ‘s de novo review to denials based on interpretations of plan terms. Our outlier view did not affect a great number of ERISA cases, however, because delegation clauses that remove a case from the default standard of are so prevalent. But the importance of this issue may be growing. As part of a trend in a number of states,1 Texas recently enacted a law banning insurers’ use of delegation clauses. TEX. INS. CODE § 1701.062(a). Assuming that the antidelegation statute is not preempted by federal law—something we do not decide today as that defense has not been asserted—a lot more ERISA cases will be subject to ‘s default standard of review. So we granted en banc review of this case to reconsider and determine the default standard of review that applies when a beneficiary challenges a plan denial based on a factual determination of ineligibility.

I.

Ariana M. is a dependent covered by an Eyesys Vision Inc. group health plan. Humana Health Plan of Texas, Inc. insures and makes benefits determinations for that plan. So when Ariana was admitted to Avalon Hills, a facility that treats eating disorders, Humana determined whether and for how long to cover her partial hospitalization. According to the plan‘s terms, partial hospitalization includes comprehensive treatment fоr a minimum of five hours per day, five days a week. This treatment is more intensive than any form of outpatient care.

When she was admitted, Ariana had over 100 self-inflicted cuts on her body, while her escalating eating disorder interfered with her ability to lead a normal life. This was no isolated occurrence. By that time, Ariana had a six-year history of eating disorders, though she claimed that her body-image dissatisfaction dated back to early childhood.

A beneficiary is only eligible for partial hospitalization for mental health services if the treatment is “medically necessary.” Medically necessary services are those “that a health care practitioner exercising prudent clinical judgment would provide to his or her patient for the purpose of preventing, evaluating, diagnosing or treating an illness or bodily injury, or its symptoms.”

Ariana‘s treatment lasted from April to September 2013. Though Humana, at various points, denied certification for continued treatment—reversing course only on appeal by Avalon Hills—it did eventually authorize forty-nine days of partial hospitalization. But Humana declined to allow partial hospitalization beyond June 5th, claiming it was no longer medically necessary.

In reaching this conclusion, Humana had two doctors evaluate Ariana‘s records. Dr. Manjeshwar Prabhu—a contract physician with Humana‘s behavioral-health vendor—conducted the initial review, finding that Ariana no longer qualified for treatment under the Mihalik criteria. Mihalik provides a set of privately licensed guidelines used to evaluate the need for certain medical services. In Prabhu‘s view, Ariana posed no imminent danger to herself or others and showed no medical instability or functional impairments, so a lower level of care, such as an intensive outpatient treatment, was appropriate. Though Avalon Hills—whose physicians participated in a peer-to-peer review of Ariana‘s case with Prabhu—acknowledged she was neither suicidal nor psychotic, it informed Prabhu that Ariana was not progressing in her treatment. In the view of a therapist at the facility, Ariana appeared to be at her “baseline behaviors.”

Avalon Hills appealed the denial. That prompted Humana to seek an additional review from Dr. Neil Hartman, a psychiatrist with Advanced Medical Reviews. He evaluated Ariana‘s medical records—including Prabhu‘s determination—and consulted her treating physicians. Hartman concluded that Ariana‘s partial hospitalization was no longer necessary because she was “medically stable,” “not aggressive,” and “not a danger to [herself or others].”

Ariana then filed this lawsuit. The plan has a clause granting to Humana “full and exclusive discretionary authority to: [i]nterpret plan provisions; [m]ake decisions regarding eligibility for coverage and benefits; and [r]esolve factual questions relating to coverage and benefits.” Early in the lawsuit, Ariana argued that the clause was unenforceable because Texas prohibits discretionary clauses. TEX. INS. CODE § 1701.062(a). In response, Humana agreed to not rely on the delegation clause (and thus did not raise a preemption defense to the Texas statute) and said it would defend its denial under the default ”de novo” standard. Despite using the ”de novo” label, Humana made clear that it was invoking the “abuse of discretion” standard applies to factual determinations even when a plan does not grant the administrator discretion. Ariana argued that the Texas law did not just invalidate delegation clauses but also overrode ‘s deferential standard of review.

The district court disagreed that Texas law could dictate the ERISA standard of review. The court thus applied and assessed whether Humana‘s decision fell “somewhere on a continuum of reasonableness—even if on the low end.” (quoting ). It held that Humana did not abuse its discretion in finding Ariana‘s continued partial hospitalization medically unnecessary—Prabhu and Hartman both conducted peer-to-peer reviews with her treating physicians, reviewed her medical files, provided reports citing the Mihalik criteria, and exрlained why she did not qualify for continued partial hospitalization under the plan. . As a result, the district court granted Humana‘s motion for summary judgment and denied Ariana‘s. .

A panel of this court affirmed. . The panel rejected Ariana‘s contention that the Texas statute mandated a specific standard of review, finding instead that the “plain text of the statute provides only that a discretionary clause cannot be written into an insurance policy.” . Therefore, Texas‘s antidelegation law did not alter “normal deference.” The panel also recognized that deference, under this court‘s long-held view, dictated abuse of discretion as the appropriate standard to review an administrator‘s factual determinations, irrespective of whether the ERISA plan contains a discretionary clause. (citing and ).

But the entire panel joined a concurring opinion questioning ‘s continuing vitality given that every other circuit to consider the standard of review issue has decided otherwise. (Costa, J., specially concurring). A number of amici, including the Department of Labor and the Texas Department of Insurance, supported Ariana‘s request for full court reconsideration of . We granted the petition.

II.

We first consider Ariana‘s argument that the Texas statute dictates the standard of review for ERISA cases. That is not our reading of the antidelegation law. It provides that an “insurer may not use a document described by Section 1701.002“—which includes health insuranсe policies—“in this state if the document contains a discretionary clause.” TEX. INS. CODE § 1701.062(a). In turn, the law defines discretionary clauses to encompass any provision that “purports or acts to bind the claimant to, or grant deference in subsequent proceedings to, adverse eligibility or claim decisions or policy interpretations by the insurer” or “specifies . . . a standard of review in any appeal process that gives deference to the original claim decision or provides standards of interpretation or review that are inconsistent with the laws of this state, including the common law.” TEX. INS. CODE § 1701.062(b)(1), (2)(D).

The Texas insurance code provision thus only renders discretionary clauses unenforceable; it does not attempt to prescribe the standard of review for federal courts deciding ERISA cases. As to whether federal law preempts this state action making discretionary clauses unenforceable, we do not consider that defense because Humana did not assert it.2

III.

With the delegation clause out of the picture and federal ERISA law providing the standard of review, this case presents us with an opportunity to reconsider . It held that “for factual determinations under ERISA plans, the abuse of discretion standard of review is the appropriate standard; that is, federal courts owe due deference tо an administrator‘s factual conclusions that reflect a reasonable and impartial judgment.” . No other circuit agrees that ‘s default de novo standard is limited to the construing of plan terms. See ; ; ; ; ; ; ; (all applying de novo review when the plan does not grant discretion).3

All but one of those courts of appeals had the opportunity to consider , and all that did so rejected its reasoning. They cited a number of reasons for not following our view. At the most basic level, they disagreed with ‘s reading of . That Supreme Court decision addressed a dispute about plan interpretation rather than one involving a factual determination that a beneficiary was not entitled to benefits. But every other circuit has read its holding as applying to both situations. That is because holds that “a denial of benefits challenged under § 1132(a)(1)(B) is to be reviewed under a de novo standard unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan.” . The first part of this pronouncement—“a denial of benefits“—does not distinguish denials that rest on contractual interpretation from those based on a factual assessment of eligibility; any denial is “to be reviewed under a de novo standard.” The end of the sentence does make that distinction in excepting from de novo review denials when plans delegate “discretionary authority to determine eligibility for benefits.”4 Why would a discretionary clause be needed for that type of decision to escape de novo review if eligibility determinations were not subject to that standard of review as a default matter? See . And while еligibility determinations may of course turn on plan interpretations, in differentiating between the two types of denials seemed to view eligibility determinations as encompassing more than just “constru[ing] the terms of the plan.” ; see (explaining that “strongly suggests that the Court intended de novo review to be mandatory where administrators were not granted discretion, regardless of whether the denials under review were based on plan interpretations” because otherwise “the Court could simply have omitted the words ‘to determine eligibility for benefits‘” (quoting )); see also (noting that benefits eligibility determinations require administrators to “determine both the facts underlying claims and whether those facts entitle claimants to benefits under the terms of the plan“).

As support for cabining de novo review only to plan interpretation, our court cited a reference early in to “actions challenging denials of benefits based on plan [term] interpretations.” (quoting ). Immediately following this language, however, the Court said it “express[ed] no view as to the appropriate standard of review for actions under other remedial provisions of ERISA.” . This suggests was articulating a general default standard of review for Section 1132(a)(1)(B) actions—the provision that allows judicial review of benefit denials—rather than making the fine distinction saw between the review of factual determinations and legal interpretations. See .

In addition to parsing the language used in , courts rejecting have noted the Supreme Court‘s observation that reading ERISA to provide a default standard of deference would undermine congressional intent as it “would afford less protection to employees and their beneficiaries than they enjoyed before ERISA was enacted.” . That concern, especially as it is imbued with concerns about the conflicts that administrators sometimes have, would not seem to be greater for legal interpretation than for factual ones. ; .

Other courts have also questioned the support found in trust law for its factual/legal dichotomy. reasoned that an administrator‘s factual determinations are inherently discretionary, in contrast to legal interpretations. It thus concluded that the Restatement (Second) of Trusts supports giving deference to an ERISA plan administrator‘s resolution of factual disputes even when the plan does not grant discretion. See (citing ).5 In a thorough examination of trust law, the Seventh Circuit disagreed with ‘s assessment. It recognized that likely flipped the presumption of trust law, which traditionally assumes deference unless the trust says otherwise.6 . But it found no trust law principles that distinguish between factual and legal determinations, as does. It concluded that the critical trust law distinction for the scope of judicial review is between powers a trust document makes discretionary and those it makes mandatory. ; see also (noting that nothing in ERISA “requires that these kinds of decisions be so ‘discretionary’ in the first place” and “whether they are is simply a matter of plan design or the drafting of an HMO contract“). To illustrate why factual determinations do not always fall on the discretionary side of that divide, points out that equity courts hаve long applied nondeferential review to a “host of factually specific decisions including

reviews of accounts and investment decisions.” ; see also (noting that the Restatement cited does not distinguish between factual and legal determinations nor have “courts reviewing the actions of trustees“).

‘s analogy to the deference that reviewing courts afford agency decisions and a district court‘s factfinding has also been criticized. One reason courts have ‍​‌‌​​‌‌‌‌​‌​​‌‌‌‌‌​​​​‌​‌‌‌​‌‌​‌‌‌‌​​‌​‌‌​​‌​​‌‌‍found the comparison inapt is that agencies and trial judges are required to apply a developed set of constitutional and statutory procedural protections. . They are also impartial whereas a plan administrator often has an incentive to reach decisions “advantageous to its own interests.” (quoting ); see also (noting that for both factual and legal determinations made by agencies, the Administrative Procedure Act requires de novo review when procedural safeguards are lacking); cf. Langbein, Trust Law as Regulatory Law, at 1326 (explaining that ERISA law differs from trust law in the “crucial respect” that “[t]rust law presupposes that the trustee who administers a trust will be disinterested, in the sense of having no personal stake in the trust assets“). Indeed, an entire body of case law has arisen to address this concern about conflicts in ERISA law, as a conflict can influence the degree of deference afforded a plan even when it is granted discretionary authority. See (requiring that district courts “take account” of conflicts in evaluating benefits denials, giving them more weight when “circumstances suggest a higher likelihood that [the conflict] affected the benefits decision“).

The passage of time has cast doubt on another reason cited for giving deference: its prediction that de novo review of factual determinations would result in a vast number of trials that would burden courts and reduce the funds available to pay legitimate claims. . But we no longer have to guess about the impact of de novo review as eight circuits have surpassed, or are nearing, two decades of experience under that regime. There is no indication that ERISA trials have depleted plan funds or overrun courts in those circuits, which are still able to grant summary judgment when the record warrants it. See, e.g., (affirming the district court‘s grant of summary judgment after the district court conducted full de novo review of the administrator‘s disability benefits denial).

And the interest in efficiency is not exclusively on the side of ‘s bifurcated system of review. Abuse-of-discretion cases frequently result in litigation about the existence and extent of a conflict of interest,7 which is one of the rare areas in which a plaintiff can often expand the administrative record with discovery. See (explaining that our restrictive position on adding to the administrative record in ERISA cases does not prohibit a discovery request for information regarding the existence and extent of a conflict). Conditioning deference on whether a decision is characterized as legal or factual makes ERISA another victim of the “delusive simplicity of the distinction between questions of law and questions of fact [that] has been found a will-of-the-wisp by travelers approaching it from several directions.” Nathan Isaacs, The Law and the Facts, 22 COLUM. L. REV. 1, 1 (1922); see also (recognizing that “[a]s a practical matter, factual findings and plan interpretations are often intertwined” and predicting that if review were bifurcated at the district court, there would be an “unnecessary cascade of

litigation over whether an administrator‘s action was a plan interpretation or a factual determination“).

There is thus no evidence that joining the eight other circuits that have long applied de novo review to factual determinations will create an overwhelming burden on district courts even if that concern can override the “ready access to the Federal courts” that ERISA provides. 29 U.S.C. § 1001(b); see (concluding that “the threat of increased litigation is not sufficient to outweigh the reasons for a de novo standard that we have already explained“). Moreover, as will be discussed, we maintain our precedent that largely limits judicial review to the record before the administrator, which mitigates concerns about the time and expense of litigation under a de novo standard.

In the years since all these circuits have disagreed with , the Supreme Court has decided more ERISA cases. Although none has directly confronted our issue (and thus they have not served as a basis to reconsider absent en banc review), two indicate that there is no fаct/law distinction for applying the default de novo standard. addresses how to assess conflicts of interest for plans that give administrators discretion. See . Humana and the dissent emphasize its comment about not wanting to “overturn by adopting a rule that in practice could bring about near universal review by judges de novo—i.e., without deference—of the lion‘s share of ERISA plan claims denials.” . But that statement discussed the prospect of de novo review for plans that validly confer discretion on administrators. . That is not at issue here. Relevant to our question about the default standard of review is ‘s list of background ERISA principles in the beginning of the opinion. Number “2” reaffirms ‘s reading of trust law and the default standard of review: “Principles of trust law require courts to review a denial of plan benefits ‘under a de novo standard’ unless the plan provides to the contrary.” (quoting ). As in , the language broadly speaks of “a denial of plan benefits” without differentiating based on the nature of the denial.

The preemption decision in also supports the broader interpretation of ‘s de novo review. . held that an Illinois law requiring independent medical review of certain benefit denials was not preempted. . That state law required independent evaluations for, among other things, the medical necessity determinations also made in this case. . The court rejected a preemption defense because ERISA does not provide a statutory standard of review. It then explained—in the context of assessing a statute that applies to factbound medical necessity determinations—that when filled in that statutory gap it “held that a general or default rule of de novo review could bе replaced by deferential review if the ERISA plan itself provided that the plan‘s benefit determinations were matters of high or unfettered discretion.” (citing ). Again, the reference is to “benefit determinations” with no distinction for legal or factual rulings. And the Court went on to say that nothing in ERISA “requires that these kinds of decisions be so ‘discretionary’ in the first place” and “whether they are is simply a matter of plan design or the drafting of an HMO contract.” . thus recognizes and analyzes the dichotomy only on discretionary/nondiscretionary grounds, not factual/legal ones. It also is yet another Supreme Court rejection of the notion that ERISA administrators are inherently entitled to discretion (even if that is what trust law provides).

Considering these cases and without having to endorse all the critiques other circuits have made of , on balance we conclude that they warrant changing course and adopting the majority approach—an approach the federal and Texas governments also support. We are also influenced by ERISA‘s strong interest in uniformity. See . Being on the lonely side of the lopsided split means that ERISA denials involving nondiscretionary plans are reviewed with more deference in Texas, Louisiana, and Mississippi than they are in the rest of the country. It even means that employees working for the same company with the same health or retirement plan mаy suffer different fates in court depending on the circuit where they reside.8 Although sometimes there is virtue in being a lonely voice in the wilderness, in this instance we conclude that one really is the loneliest number. See Three Dog Night, One, on THREE DOG NIGHT (Dunhill 1969). We overrule and now hold that ‘s default de novo standard applies when the denial is based on a factual determination.

IV.

Changing the standard of review does not require us to alter our precedent concerning the scope of the record in ERISA cases. Although other circuits are unanimous on what the default standard of review is, they take a variety of positions on whether de novo review allows a party to expand the record beyond what was before the plan administrator. Some do not limit reviewing courts to that record. See (holding that limiting the judicial record to that before the plan administrator is not appropriate in de novo cases); (finding that limiting a district court to the record before a plan administrator “makes little sense” because it is contrary to the ordinary concept of de novo review). Others take a more restrictive view. See (admonishing district courts to avoid admitting additional evidence “absent good cause to do so“); (permitting district courts to admit additional evidence only in “necessary” and “[e]xceptional circumstances“).

Our leading case in this area is (en banc), overruled on other grounds by . Under , a plan administrator must identify evidence in the administrative record, giving claimants a chance to contest whether that record is complete. . Once the recоrd is finalized, a district court must remain within its bounds in conducting a review of the administrator‘s findings, even in the face of disputed facts. permits departure from this rule only in very limited circumstances. One exception allows a district court to admit evidence to explain how the administrator has interpreted the plan‘s terms in previous instances. (citing ). Another allows a district court to admit evidence, including expert opinions, to assist in the understanding of medical terminology related to a benefits claim. Those situations are not actually expanding the evidence on which the merits are evaluated but providing context to help the court evaluate the administrative record.

Although some of ‘s reasoning for limiting the district court record to what was before the administrator depended on the abuse-of-discretion context, other interests it recognized support the same rule for de novo review. Among those is the interest in encouraging parties to resolve their dispute at the administrative stage. . A different standard of review also does not undermine ‘s observation that there is not a “particularly high bar to a party‘s seeking to introduce evidence into the administrative record.” And generally limiting the evidence to what was in front of the plan administrator when a dispute ends up in court allows for speedier resolution.

In short, overruling while adhering to in the context of de novo review serves the twin ERISA goals of allowing for efficient yet meaningful judicial review. See 29 U.S.C. § 1001(b) (stаting that ERISA is intended to provide “ready access to the Federal courts“); (explaining that a deferential default standard “would afford less protection to employees and their beneficiaries than they enjoyed before ERISA was enacted“). will continue to provide the guiding principles on the scope of the record for future cases that apply de novo review to fact-based benefit denials.

V.

This brings us back to Ariana‘s claim. Following , the district court concluded only that “Humana did not abuse its discretion in finding that Ariana M.‘s continued treatment at Avalon Hills was not medically necessary after June 4, 2013.” . That determination is now subject to de novo review. A different standard of review will sometimes lead to a different outcome, but there will also be many cases in which the result would be the same with deference or without it. We give no opinion on which is the case here, but leave application of the de novo standard to the able district court in the first instance.9

*

*

*

The judgment of the district court is VACATED and REMANDED for further proceedings consistent with this opinion.

E. GRADY JOLLY, Circuit Judge, dissenting, joined by JONES, SMITH, CLEMENT, and ELROD, Circuit Judges. OWEN, Circuit Judge, joins only Part I.

The material question in this en banc appeal is whether , requires de novo review of an administrator‘s entitlement determinations; that is to say, whether an administrator‘s findings of fact underlying the merit of a participant‘s claim are entitled to any deference by the federal courts. Read holistically—that is, by considering the context in which came before the Supreme Court; the Court‘s opinion as a whole, instead of snippet by snippet; and the Supreme Court‘s concerns that it expressed during ‘s oral argument— speaks to de novo review in relation to eligibility determinations and the construction of plan terms—both inherently legal questions—not to the daily grind of winnowing the merit of individual factual claims.

Moreover, the majority opinion reflects an impractical view of the administrative process. It is inconsistent with the law of trusts and misreads subsequent cases decided by the United States Supreme Court. I respectfully dissent.

I.

A.

A holistic reading of makes clear that its de novo standard of review applies only to legal questions.

The first step of understanding requires examining the facts and law that were asserted in the courts below; that is, determining what sort of case was actually before the Supreme Court. involved the construction of plan terms under three different ‍​‌‌​​‌‌‌‌​‌​​‌‌‌‌‌​​​​‌​‌‌‌​‌‌​‌‌‌‌​​‌​‌‌​​‌​​‌‌‍ERISA benefits plans maintained by Firestone. . Firestone was the administrator

and the defendant. Firestone had construed the terms of its ERISA plans to deny severance benefits to “former employees” who worked at Firestone plants that had been sold to another company. . The former Firestone employees sued, disputing Firestone‘s interpretation of the plan as to whether they were eligible for benefits under the terms of the plan. . The district court granted Firestone summary judgment, holding that Firestone‘s decision to deny benefits was not arbitrary and capricious. . The court of appeals reversed on grounds that Firestone, as administrator of the plan, had a conflict of interest; as such, de novo—not “arbitrary and capricious“—was the proper standard of review for Firestone‘s interpretation of the plan. . Thus, when the case reached the Supreme Court, the appeal was twofold: First, whether Firestone‘s interpretation of the plan was subject to an arbitrary-and-capricious standard of review or to a de novo review and, second, whether a proper interpretation of the plan‘s terms covered these “former employees.” The Supreme Court did not address whether the courts should defer to administrators’ factual decisions because the court of appeals had reserved comment on that question. See (“It should be noted that we also do not deal here with a determination of fact by a plan administrator. We leave for another day the definition of the context, if any, in which courts should defer to such a determination.“), cited in . Thus, it should be clear that the “denial of benefits” before the Supreme Court was the denial of benefits to a class of participants based on Firestone‘s interpretation of the plan as to that class, not a denial of the underlying merit of a participant‘s claim.

The second step of understanding the Firestone opinion requires the opinion be read as a whole and in context. The Supreme Court sets the tone of its analysis when it limits its holding to disputes based on interpretation of the plan; indeed, as noted above, these legal questions were the only issues before the Court. The Firestone Court said, “The discussion which follows is limited to the appropriate standard of review in § 1132(a)(1)(B) actions challenging denials of benefits based on plan interpretations. We express no view as to the appropriate standard of review for actions under other remedial provisions of ERISA.” (emphasis added). The Court then said, ”ERISA does not set out the appropriate standard of review for actions under § 1132(a)(1)(B) challenging benefit eligibility determinations,” (emphasis added), signaling again that it was addressing an inherently legal question. The Court then rejected the arbitrary-and-capricious standard of review (applied by the district court), saying “the wholesale importation of the arbitrary and capricious standard into ERISA [was] unwarranted.” In doing so, the Court plainly did not suggest that a uniform standard of review applied to all decisions of the administrator. Further into its opinion, the Supreme Court clarified what it meant by “plan interpretations,” saying,

As this case aptly demonstrates, the validity of a claim to benefits under an ERISA plan is likely to turn on the interpretation of terms in the plan at issue. Consistent with established principles of trust law, we hold that a denial of benefits challenged under § 1132(a)(1)(B) is to be reviewed under a de novo standard unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan.

(emphasis added).1 Plan-term construction and eligibility determinations are both legal concepts that are part of “plan interpretation.” Neither concept addresses whether, under the undisputed provisions of the plan, a specific person‘s individual claim has merit.

The third step of understanding the holding of the Firestone Court involves the transcript of the oral argument before the Supreme Court. The oral argument confirms that the subject before the Court was plan interpretation. There, several Justices stated the issue in terms of whether the Court must give deference to ERISA plan administrators for their construction of plan terms. See Oral Argument at 3:05, Firestone Tire & Rubber Co. v. Bruch, (No. 87-1054), https://www.oyez.org/cases/1988/87-1054 (Justice White saying to Firestone‘s lawyer that “it‘s a contractual construction problem. Or it‘s a construction of a written instrument“); at 4:35 (Justice Scalia saying to Firestone‘s lawyer that “it‘s not my impression of common law trust law that if the trustee makes a questionable interpretation of the trust agreement, I wouldn‘t be able as one of the beneficiaries to go into court and say that interpretation is wrong. And the court would look at the trust agreement and say it‘s up to us to interpret this trust agreement“). Further, a Justice expressed that the case might be different if resolving a question involving a factual judgment, by saying:

My, my recollection of trust law . . . and it obviously isn‘t, isn‘t a terribly recent one . . . is that if you‘re talking about the . . . the many things that the trustee is given discretion to do in a trust instrument, decide on the medical needs or educational needs of various beneficiaries and allocate discretionary funds among them, the courts give great deference to a trustee.

But is . . . in deciding who is a beneficiary, I, I was not aware that trust law says the trustee has great discretion therе.

at 5:23. And the plaintiffs’ lawyer emphasized that this case involved only “a pure question of plan interpretation” and involved a different “category of question” from a fact question. at 32:49, 36:33.2 Nothing in the argument signals that the Court considered that its ruling, i.e., applying de novo review to who is a beneficiary under the plan, would also apply to fact questions.

Therefore, based on the procedural history, the proper context, the oral argument, and the specific language of the opinion, it should be clear to all but the obstinate that the Firestone Court did not intend that de novo review would apply to factual questions that went before plan administrators.

B.

The majority‘s argument that Firestone mandates de novo review for factual issues is further undermined by Firestone‘s clarity that principles of trust law apply to administrator actions. See (“In determining the appropriate standard of review for actions under § 1132(a)(1)(B), we are guided by principles of trust law.“). The majority seems to disregard this directive. But, inasmuch as Firestone clearly does not mandate a de novo standard of review for factual disputes, trust law controls, as instructed by Firestone.

Under trust law, trustees have measured discretion in determinations that fulfill the underlying purposes of the trust; yet, the majority, with its de novo review, grants trustees no deference in administering the quotidian claims arising under the trust document. The Second Restatement provides that trust administrators have two types of powers: (1) those conferred upon the administrator “in specific words by the terms of the trust” and (2) those “necessary or appropriate to carry out the purposes of the trust and are not forbidden by the terms of the trust.” Restatement (Second) of Trusts § 186 (Am. Law Inst. 1959). The Third Restatement explains further, “When a trustee has discretion with respect to the exercise of a power, its exercise is subject to supervision by a court only to prevent abuse of discretion.” Restatement (Third) of Trusts § 87 (Am. Law Inst. 2007). The general rule is that trustees have discretion with respect to the exercise of trusteeship powers, except when directed differently by the terms of the trust or when compelled by the trustee‘s fiduciary duties. Id. cmt. a.

And turning to Scott and Ascher on Trusts, we find, “A trustee‘s powers ordinarily are discretionary, unless the terms of the trust or applicable law makes them mandatory.” 3 A. Scott & M. Ascher, Scott and Ascher on Trusts § 18.2, p. 1338 (5th ed. 2007). Trustees have “considerable discretion in determining what is necessary for any given beneficiary‘s support,” and courts ensure only that trustees do not exceed the limits of their discretion. Id. § 18.2.6, at 1362. Indeed, “[t]he court will not substitute its own judgment for that of the trustee” because “[t]he mere fact that the court would have exercised the power differently is not a sufficient reason for the court to interfere.” Id. § 18.2, at 1340. Instead, the court may сheck the trustee‘s powers by examining whether the trustee (1) abused its discretion, (2) acted dishonestly or in bad faith, and (3) exercised its reasonable judgment when exercising its powers. Id. § 18.2.2–18.2.6, at 1350–67.

As we have earlier noted, the Firestone Court expressly said that its decision was guided by principles of trust law. Here, whether a covered beneficiary has presented facts to support the benefits she individually claims is a core discretionary power that is “necessary or appropriate” to the routine administration of plans. As we said in Pierre, “[i]t is indisputable that an ERISA trustee, by its very nature, is granted some inherent discretion, i.e., ‘authority to control and manage the operation and administration of the plan.‘” (quoting 29 U.S.C. § 1102(a)(1)).3 The majority would not allow a smidgeon of deference to the administrator, a position that is contrary to the guiding advice of Firestone.

II.

We leave Firestone proper for a moment and turn our attention to recent Supreme Court cases also dealing with the administration of ERISA plans. In particular, two recent Supreme Court opinions strongly support that the Supreme Court would conclude that Pierre correctly states the law: , and .4 Conkright—not cited by the majority—reaffirmed that in § 1132(a)(1)(B) cases, we should look to the principles of trust law. See (“In determining the proper standard of review when a plan administrator operates under a conflict of interest [in Glenn], we again looked to trust law, the terms of the plan at issue, and the principles of ERISA—plus, of course, our precedent in Firestone.“).

The Supreme Court‘s opinion in Glenn supports Pierre‘s understanding of Firestone. In Glenn, the Court said,

We do not believe that Firestone‘s statement implies a change in the standard of review, say, from deferential to de novo review. Trust law continues to aрply 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. We see no reason to forsake Firestone‘s reliance upon trust law in this respect.

(internal citations omitted) (citing ; Restatement § 187, cmts. d-j; Scott and Ascher on Trusts § 18.2, at 1342-44). The Court then emphasized that it would not “overturn Firestone by adopting a rule that in practice could bring about near universal review by judges de novo—i.e., without deference—of the lion‘s share of ERISA plan claims denials,” because it believed that Congress would have said more about such a standard of review if it wanted the courts to have wholesale review. . The Glenn Court quoted Justice Scalia‘s pithy and colorful admonition that “Congress does not ‘hide elephants in mouseholes,‘” (quoting ), i.e., if Congress had intended a radical departure from traditional principles of trust law, it most certainly would have not hidden it in statutory interstices.

Thus, the majority‘s view brushes aside the admonition of Glenn that Firestone cannot be read to endorse “near universal review” of all plan denials brought to our district courts. Other circuits may have interpreted Firestone in their own way fifteen to twenty years ago, but, today, it should be understood that, in the light of more recent Supreme Court cases, Firestone did not change ERISA‘s application of trust law.

These Supreme Court cases (each decided after the decisions of the other circuit courts to the contrary) further undermine the rationale offеred by the majority to strip the administrator of discretionary respect. Other circuits, and now the majority, have acknowledged that federal courts are required generally to pay deference to administrative decisions. But, the majority argument goes, plan administrators do not have the expertise of administrative agencies, and ERISA administrators are not unbiased factfinders. See Maj. Op. at 11; ; ; . It follows, says the majority, that the usual deference to administrators is not warranted for ERISA administration. See Maj. Op. at 11.

Never mind that this concern, too, was addressed by Glenn. The Supreme Court favorably compared ERISA‘s review of benefits decisions to review of administrative agencies’ decisions by observing,

This kind of review is no stranger to the judicial system. Not only trust law, but also administrative law, can ask judges to determine lawfulness by taking account of several different, often case-specific, factors, reaching a result by weighing all together.

. For this statement, the Court cited two administrative law decisions— and —in which the Supreme Court reviewed a governmental decision and an agency‘s factfindings for abuse of discretion.5 And Glenn itself dealt with the biggest concern arising from plan administrators—conflicts of interest—by instructing that whenever a district court reviews a plan administrator for abuse of discretion, that court must consider the extent of any conflict. .6 Conkright, although discussing a plan with a clause that provided the administrator with discretionary review, similarly endorsed providing deference to ERISA plan administrators because the practice “promotes efficiency by encouraging resolution of benefits disputes through internal administrative proceedings rather than costly litigation.” .

III.

The majority‘s moving force for overruling Pierre is that we should join the other circuits because ERISA must be uniformly applied among the federal circuit and district courts. Indeed, the Supreme Court in Conkright allowed: ”ERISA ‘induc[es] employers to offer benefits by assuring a predictable set of liabilities, under uniform standards of primary conduct and a uniform regime of ultimate remedial orders and awards when a violation has occurred.‘” (alteration in original) (quoting ).

But the other circuits, which declined to follow Pierre, and which the majority would have us reverse our course and follow, are outdated by Glenn and Conkright.7 And still further, the uniformity ‍​‌‌​​‌‌‌‌​‌​​‌‌‌‌‌​​​​‌​‌‌‌​‌‌​‌‌‌‌​​‌​‌‌​​‌​​‌‌‍that might result from reversing Pierre is illusory. First, different circuits have different standards for reviewing evidence. Some circuits pay little or no attention to the administrative record and virtually allow trial de novo by opening discovery in district court. See (“We agree with the Third, Fourth, Seventh, Eighth, and Eleventh Circuits that new evidence may be considered under certain circumstances to enable the full exercise of informed and independent judgment.“). Other circuits, like ours, are limited to the record that the administrator considered. See Maj. Op. at 17 (limiting district court proceedings to the administrative record);8 (preventing district courts from considering evidence outside the record). Second, some states have anti-discretionary-clause statutеs—like Texas Insurance Code § 1701.062(a) that the defendants decided not to challenge here—that do not allow plans to grant discretionary authority to plan administrators, even though the Supreme Court has relied upon discretionary clauses and approved of them multiple times. See (stating that deference to plan administrators promotes efficiency, predictability, and uniformity); (approving the use of discretionary clauses because “participants are not likely to value judicial review of plan determinations over internal review“). Thus, a claimant—in a case involving an ERISA plan with a discretionary clause—will have a different standard of review depending on whether she brings an action in a state in which she resides or a state in which a breach occurred. See 29 U.S.C. § 1132(e)(2) (allowing ERISA suits to proceed in any federal district court “where the plan is administered, where the breach took place, or where a defendant resides or may be found“). For example, an ERISA plan enforced in Louisiana, which does not have a state statute prohibiting discretionary clauses, will have a different standard of review than if it were enforced in Texas, which prohibits discretionary clauses.

If uniformity were the Holy Grail to be pursued among federal courts and if the instant opinion accomplished uniformity, the majority opinion would be more рersuasive. But, although I agree that ERISA uniformity is a worthy consideration, the majority‘s opinion hardly establishes greater procedural (or substantive) uniformity than if we continued to apply Pierre. Instead, we are left with a strained argument for uniformity and the illusion that reversing Pierre somehow accomplishes uniformity throughout the federal courts of the country.

IV.

To sum up: the misguided majority upsets twenty-six years of precedent in overruling Pierre, and for no compelling reason. In doing so, it ignores the practicality of administrative and trust law, misreads Firestone, and is swept up by outdated cases of other circuits. Respectfully, I dissent.9

PRISCILLA R. OWEN, Circuit Judge, dissenting:

The Supreme Court has not decided whether a de novo or an abuse of discretion standard of review applies when an ERISA plan administrator considers conflicting expert opinions and denies coverage for the continued hospitalization of an ERISA welfare-plan beneficiary. However, if the principles of trust law are applied, as the Supreme Court has repeatedly said that they should be, then an abuse of discretion standard is applicable in the present case. I would therefore affirm the judgment of the district court.

Ariana M. brought the present action under 29 U.S.C. § 1132(a)(1)(B).1 The Supreme Court explained in that “[i]n determining the appropriate standard of review for actions under § 1132(a)(1)(B), we are guided by principles of trust law.”3 The only issue before the Supreme Court in Firestone was the standard of review that should apply to a plan administrator‘s interpretation of the plan.4 The Court held that a de novo standard of review applied, explaining, in part, that “[t]he trust law de novo standard of review is consistent with the judicial interpretation of employee benefit plans prior to the enactment of ERISA.”5 The Court reasoned that “[a]ctions challenging an employer‘s denial of benefits before the enactment of ERISA were governed by principles of contract law,” and that “[i]f the plan did not give the employer or administrator discretionary or final authority to construe uncertain terms, the court reviewed the employee‘s claim as it would have any other contract claim—by looking to the terms of the plan and other manifestations of the parties’ intent.”6 But the Court looked primarily to the law governing trusts in reaching its decision.

In Firestone, the Court considered the Restatement (Second) of Trusts (1959) (hereinafter “the Restatement“), which was the current version of the Restatement of Trusts at the time of ERISA‘s enactment.7 The actual holding in Firestone was entirely consistent with Section 201, comment b, of the Restatement, which provides that “[t]he extent [of a trustee‘s] duties and powers is determined by the trust instrument and the rules of law which are applicable, and not by his own interpretation of the instrument or his own belief as to the rules of law.”8 Accordingly, under the Restatement, unless a trust instrument provides to the contrary, a trustee‘s interpretation of the terms of the trust would be subject to de novo review by a court. The Supreme Cоurt‘s holding in Firestone that an abuse of discretion standard should be applied to an ERISA administrator‘s interpretation of the plan only when the plan grants discretion to the administrator to interpret the plan is in line with Section 201, comment b.9

However, the Restatement makes clear in Section 187, comment a, that “except to the extent to which its exercise is required by the terms of the trust or by the principles of law applicable to the duties of trustees,” a trustee‘s “exercise of power is discretionary.”10 Other comments in Section 187 of the Restatement support the conclusion that a trustee‘s decision as to whether a beneficiary‘s condition entitles her to benefits from the trust is within the trustee‘s discretion. Comment c provides that a trustee has discretion “to determine the amount necessary for a beneficiary‘s support.”11 The Restatement makes clear that when a power is committed to the discretion of a trustee, his actions or inactions are to be judged by an abuse of discretion standard. Comment d sets forth the “[f]actors in determining whether there is an abuse of discretion.”12 Comment e explains when there is “[n]o abuse of discretion.”13

Ariana M.‘s claim that she was entitled to payment for continued hospitalization as a beneficiary under an ERISA welfare benefits plan necessarily involves the exercise of judgment by the ERISA plan administrator in analyzing cоnflicting expert opinions. This is the type of decision that would be committed to the discretion of a trustee under trust law, as expressed in the Restatement.

Though the Supreme Court has spoken in broad terms when it has said that a de novo standard of review applies to a court‘s review of the “denial of [ERISA] plan benefits”14 unless the plan grants the plan administrator “discretionary authority to determine eligibility for [ERISA] benefits,”15 the Court‘s decisions have involved either a plan administrator‘s interpretation of the plan (not an administrator‘s decision as to whether, as a factual matter, the beneficiary‘s condition required a specific course of treatment),16 or a plan that expressly granted the plan administrator “discretionary authority to determine whether an employee‘s claim for benefits is valid.”17 If we are to accept the Supreme Court‘s repeated statements that principles of trust law apply when a court reviews the denial of ERISA benefits, then the determination at issue in the present case was committed to the discretion of the plan administrator, and an abuse of discretion standard should apply.

* * *

Because I would affirm the district court‘s judgment, I respectfully dissent.

JENNIFER WALKER ELROD, Circuit Judge, joined by JOLLY and CLEMENT, Circuit Judges, dissenting:

I write separately to address the decision to remand this case to the district court. This is a waste of judicial resources because there is no genuine issue of material fact and the record estаblishes that the plan administrator did not err in declining to cover Ariana‘s additional partial hospitalization. I would affirm the district court‘s judgment in favor of Humana, regardless of whether we apply the de novo standard adopted by the majority opinion today or the standard we previously adopted in .

In ERISA cases, “[w]e review a ‘district court‘s grant of summary judgment de novo, applying the same standards as the district court.‘” (quoting ). Summary judgment is appropriate when “there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(a).

There is no genuine issue of material fact on this record that precludes summary judgment. At oral argument, Ariana‘s counsel could point to only one possible area of disputed fact. Ariana‘s counsel seemed to suggest that there is a genuine dispute as to whether Dr. Hartman was qualified to make a decision about the necessity of Ariana‘s continued partial hospitalization.1 See Oral Argument at 7:25, Ariana M. v. Humana Health Plan of Tex., Inc., No. 16-20174 (5th Cir. argued Sept. 19, 2017) (en banc). When asked what evidence supported her position that Dr. Hartman was not qualified, Ariana‘s counsel referenced a deposition of Dr. Hartman. But the district court did not consider this deposition testimony “because depositions taken in earlier actions mаy only be used in a later action involving the same subject matter between the same parties.” (quoting Fed. R. Civ. P. 32(a)(8)). And Ariana did not appeal the district court‘s decision on this evidentiary issue. See (“Claims not pressed on appeal are deemed abandoned.“).

Furthermore, in accordance with (en banc), overruled on other grounds by , left intact by the majority opinion, “the district court is constrained to the evidence before the plan administrator,” even in the face of disputed facts. . Dr. Hartman‘s deposition testimony was not part of the administrative record. See . Thus, Ariana points to no evidence in the record in support of such a dispute.

In her initial brief to the panel, Ariana seemed to suggest that the fact that her doctors disagreed with the assessments of Humana‘s reviewing doctors regarding the proper level of care for Ariana created a fact issue. However, the Supreme Court has held, in an opinion issued after , that nothing in ERISA “suggests that plan administrators must accord special deference to the opinions of treating physicians. Nor does the Act impose a heightened burden of explanation on administrators when they reject a treating physician‘s opinion.” . Thus, this court need not defer to the opinions of Ariana‘s doctors, and the fact that they conflict with those of Humana‘s doctors does not create a fact issue.

For the reasons discussed in detail in the district court‘s opinion, see , the plan administrator did not err in deciding that Ariana M.‘s continued partial hospitalization was not medically necessary. As the district court explained, “Dr. Prabhu and Dr. Hartman—board-certified psychiatrists—both did peer-to-peer reviews with Ariana M.‘s health-care professionals and reviewed her medical files to apply the plan‘s terms. They set out their decisions in written reports that cited the Mihalik criteria and explained why Ariana M. failed to meet several prerequisites for continued treatment under the plan.” .

The district court‘s grant of summary judgment should be affirmed.

Notes

1
Twenty-six states, including Texas, have moved to prohibit discretionary clauses either through statute or regulatory action. Nat‘l Ass‘n of Ins. Comm‘rs, Prohibition on the Use of Discretionary Clauses Model Act ST-42-3-6 (2014), http://www.naic.org/store/free/MDL-42.pdf. Louisiana and Mississippi have not taken any such action. Id. ST-42-4. The majority mistakenly relies on this portion of the opinion to say that Firestone applies de novo review to both factual and legal assessments by plan administrators. See Maj. Op. at 8, 14. But the majority makes a mistake by conflating eligibility, i.e., coverage determinations, with entitlement determinations, i.e., claim merits, as do our sister circuits. See Maj. Op. at 8 (“Why would a discretionary clause be needed . . . to escape de novo review if eligibility determinations were not subject to that standard of review as a default matter?” (emphasis added)); (stating that the cоurt of appeals must rely on the Supreme Court‘s use of “eligibility for benefits,” which is a distinct issue from construing the plan‘s terms); (stating that “the explicit reference to ‘eligibility‘” means that the Firestone Court meant to cover entitlement decisions). This view is misguided. As evidenced by its analogy of ERISA to contract law and by its statement that ERISA was meant “to protect contractually defined benefits,” (quoting ), the Supreme Court‘s discussion of eligibility concerns whether the plan covered the person or claim at issue, not whether the covered person‘s factual context entitled her to benefits under the plan or whether the covered claim had merit. We may illustrate this distinction by considering the case at hand. Here, it is clear that Ariana M is an eligible participant and that her claim is eligible under the terms of the group health plan covering mental illness. If the administrator argued otherwise, de novo review would be used, like in any other contract dispute, to determine whether Ariana‘s claim is contractually barred. But that is not the appeal here. Instead, Ariana asks us to reevaluate the facts upon which the administrator denied the merits of her claim—that is to say, the factual claim of whether her treatment was medically necessary; such a question requires, not legal analysis, but credibility determinations, particularly among the parties’ respective experts. The majority would grant the federal courts the authority to relitigate in federal court that credibility determination, robbing the administrator of all deference to its decision. Such federal court authority does not have its source anywhere in Firestone. 29 U.S.C. § 1132(a)(1)(B) provides:

(a) Persons empowered to bring a civil action

A civil action may be brought--

(1) by a participant or beneficiary--

(B) to recover benefits due to him under the terms of his plan, to enforce his rights under the terms of the plan, or to clarify his rights to future benefits under the terms of the plan . . . .

Later during the argument, however, Ariana‘s counsel seemed to concede that there is no genuine dispute of material fact, stating that “this court could decide whether these services were primarily for the convenience of Ariana M. or were for her treatment, if this court wants to decide that . . . .” Oral Argument at 58:43, Ariana M., No. 16-20174 (5th Cir. argued Sept. 19, 2017) (en banc).
2
Each court to decide this issue has concluded that ERISA ‍​‌‌​​‌‌‌‌​‌​​‌‌‌‌‌​​​​‌​‌‌‌​‌‌​‌‌‌‌​​‌​‌‌​​‌​​‌‌‍does not preempt state antidelegation statutes. See ; ; ; see also (stating that a full ban on discretionary clauses would not likely be preempted, even though ERISA preempted a state statute regulating them). The plaintiffs also conceded in their brief that they were “not challenging the exercise of any authority which is inherently discretionary in nature.” Brief for the Respondents, (No. 87-1054), at 24. And they agreed with Firestone that courts should defer to those who have some amount of decision-making authority. .
3
(noting the court‘s task of independently weighing the facts and opinions in the administrative record and “giv[ing] no deference to the administrator‘s opinions or conclusions“). The majority sees it otherwise, adopting the Seventh Circuit‘s view that Firestone “reversed the presumption” for all plan-administrator decisions unless a plan term gives the administrator discretion. See Maj. Op. at 10; . Under that view, Firestone essentially eliminated all discretionary administrable powers—defined as those “necessary or appropriate to carry out the purposes of the trust and are not forbidden by the terms of the trust,” Restatement (Second) of Trusts § 186(b)—and permitted trustees discretionary powers only when conferred. See id. § 186(a). But this view is mistaken and departs from Firestone‘s command to use traditional trust law principles when examining § 1132(a)(1)(B). See . (citing ).
4
Judge Jolly‘s dissenting opinion contends that “eligibility for benefits” refers only to whether a person or type of claim is covered under the plan as a legal matter, not to the factual question at issue here regarding whether the plaintiff‘s claim should be paid. Dissenting Op. at 3 n.1. For starters, this ignores that says “a denial of benefits,” without qualification, is reviewed de novo, and that the “eligibility for benefits” language appears in the clause saying a grant of discretionary authority can change that standard. . So the more limited meaning of “eligibility” the dissent urges would only narrow the effect of discretionary clauses in being able to change the default de novo standard for “a denial of benefits.”

More fundamentally, the dissent‘s understanding of “eligibility” is at odds with ERISA‘s text. What the dissent describes as the initial coverage determination is a question of whether a claimant is a “participant” or “beneficiary” (Ariana is the latter as a dependent of a participant). The statute defines a “participant” as “any employee or former employee of an employer . . . who is or may become eligible to receive a benefit of any type from an employee benefit plan.” 29 U.S.C. § 1002(7) (emphasis added). So someone is covered under the plan even if they are nоt yet eligible to receive a benefit, such as someone still working who does not yet receive pension payments or someone covered under a medical policy who has not yet been to a doctor. That someone can be covered under the policy who is not yet “eligible” to receive benefits shows that an “eligibility determination” is not the same question as whether the person is covered. Our cases have long reflected this understanding in using “eligibility determination” to describe claims like this one that turn on factual entitlement to benefits. See, e.g., (evaluating whether plaintiff was “eligible” for disability benefits based on multiple doctors’ reports); (finding that plan fiduciaries are not required to obtain proof of substantial change in a plan recipient‘s medical condition after the “initial determination of eligibility” if they receive additional medical information suggesting “a covered employee” is no longer “eligible for benefits“); (labelling a factual dispute about “medical necessity” as a question of “eligibility determination“). Finally, both the majority and dissenting opinions in the Supreme Court‘s decision—a case discussed more below—treated a factual medical necessity issue as an eligibility determination. (stating that an Illinois law requiring an “independent reviewer‘s de novo examination” of medical necessity “mirrors the general or default rule we have ourselves recognized” in ); (Thomas, J., dissenting) (calling the issue “purely an eligibility decision with respect to reimbursement“).

And when faced with this precise question—whether Firestone mandates de novo review for factual entitlement decisions—the Supreme Court has denied certiorari twice in the past decade. See ; . (“The discussion which follows is limited to the appropriate standard of review in § 1132(a)(1)(B) actions challenging denials of benefits based on plan interpretations. We express no view as to the appropriate standard of review for actions under other remedial provisions of ERISA.“).
5
Section 186(b) provides that “the trustee can properly exercise such powers and only such powers as . . . are necessary or appropriate to carry out the purposes of the trust and are not forbidden by the terms of the trust.” Restatement (Second) of Trusts § 186(b). Section 187, meanwhile, states that “[w]here discretion is conferred upon the trustee with respect to the exercise of a power, its exercise is not subject to control by the court, except to prevent an abuse by the trustee of his discretion.” Id. § 187. Our precedent, too, has said that review of ERISA benefits determinations is like review of administrative agency decisions. See (“[O]ur review of an ERISA benefits determination is essentially analogous to a review of an administrative agency decision . . . .“). And with good reason: “‘[F]ull review of the motivations behind every plan administrator‘s discretionary decisions’ would ‘move toward a costly system in which Article III courts conduct wholesale reevaluations of ERISA claims’ and would seriously undermine ERISA‘s goal of resolving claims efficiently and inexpensively.” (quoting ). .
6
A leading trust scholar left no doubt of what he thought about ‘s reading of trust law. See John H. Langbein, The Supreme Court Flunks Trusts, 1990 SUP. CT. REV. 207. Professor Langbein explained that the Supreme Court reversed the traditional trust-law presumption that assumed “[t]he trustee ha[d] discretion unless the instrument or some particular doctrine of trust law denies discretion.” Id. at 219. Despite his sharp critique of the Supreme Court‘s reading of trust law, Professor Langbein believes the Court correctly adopted de novo review in light of “the regulatory purposes of ERISA.” John H. Langbein, Trust Law as Regulatory Law: The UNUM/Provident Scandal and Judicial Review of Benefit Denials Under ERISA, 101 NW. U. L. REV. 1315, 1323 n.47 (2007). Of course, regardless of whether was right or wrong in setting a default de novo standard, we are bound to apply it. The majority seems to lean heavily on conflicts of interest to justify de novo review for all decisions of administrators. See Maj. Op. at 9, 11. But conflicts of interest already must be considered as a factor in every § 1132(a)(1)(B) case, whether the standard of review is de novo or abuse of discretion, because of the requirements set out in Glenn. See . (citing cases).
7
See . The last circuit squarely to decide this issue did so 15 years ago in 2003. See ; ; ; ; ; ; . ; see also (SCALIA, J., dissenting) (noting that the Firestone decision “[c]it[ed] the Restatement of Trusts current at the time of ERISA‘s enactment“); Employment Retirement Income Security Act of 1974, Pub. L. 93-406, Title I, § 502, 88 Stat. 829 (codified as amended at 29 U.S.C. §§ 1001-1461).
8
The dissent argues that application of ERISA will not be uniform if state statutes сan nullify discretionary clauses. Deference would not be available in states with such laws; it would be available in other states. But that would be a difference rooted in the policy choices of the states—differences that are expected and honored in our federal system—and not based on inconsistent court interpretations of the same federal law. The dissent‘s argument might be relevant to a conflict preemption analysis, but as we have mentioned, we take no position on that question. I fully agree with the majority‘s decision to limit judicial review to the administrative record as we decided in (en banc), overruled on other grounds by . Restatement (Second) of Trusts § 201 cmt. b (1959), which provides:

b. Mistake of law as to existence of duties and powers. A trustee commits a breach of trust not only where he violates a duty in bad faith, or intentionally although in good faith, or negligently, but also where he violates a duty because of a mistake as to the extent of his duties and powers. This is true not only where his mistake is in regard to a rule of law, whether a statutory or common-law rule, but also where he interprets the trust instrument as authorizing him to do acts which the court determines he is not authorized by the instrument ‍​‌‌​​‌‌‌‌​‌​​‌‌‌‌‌​​​​‌​‌‌‌​‌‌​‌‌‌‌​​‌​‌‌​​‌​​‌‌‍to do. In such a case, he is not protected from liability merely because he acts in good faith, nor is he protected merely because he relies upon the advice of counsel. Compare § 297, Comment j. If he is in doubt as to the interpretation of the instrument, he can protect himself by obtaining instructions from the court. The extent of his duties and powers is determined by the trust instrument and the rules of law which are applicable, and not by his own interpretation of the instrument or his own belief as to the rules of law.

9
In light of this decision overruling our longstanding precedent and remanding for application of a de novo standard, the district court may consider whether there is good cause to allow Humana to amend its answer and assert a preemption defense if it so desires. The majority, in reference to the dissent, argues that the dissent is mistaken in its understanding of what Firestone referred to as “eligibility for benefits.” See Maj. Op. at 8–9 n.4. The majority is, of course, quite correct that our precedent has been inconsistent by using “eligibility” in some circumstances, while using “entitlement” in other circumstances, to mean determinations of factual questions. Compare (describing that power as discretion to determine “eligibility“), with (stating that the employee‘s claim failed because he did not “allege entitlement to benefits within the eligibility provision“); (stating that an employee failed to prove her entitlement when examining а plan where the employer had “discretionary authority . . . to determine eligibility for and entitlement to Plan benefits“). The relevant question, however, is not the confused use of “eligibility,” but instead what the Firestone Court meant by eligibility determinations. Given the context in which the case was decided and the language of the opinion, eligibility means qualification to claim entitlement to benefits under the plan. One may be eligible for an entitlement while not being factually entitled to the benefit. In short, eligibility precedes entitlement. One may be eligible to assert a statutory right, but only entitled to the benefits of the right upon a factual showing. Moreover, ERISA‘s text gainsays the majority‘s argument that “eligibility” and “entitlement” are fungible terms in the context of ERISA. Specifically, § 1002(7) provides that a “participant” is one “who is or may become eligible to receive a benefit.” But § 1002(7) addresses matters of coverage; that is, eligibility. On the other hand, 29 U.S.C. § 1002(8), under which Ariana M qualifies as a beneficiary, speaks in terms of entitlement to benefits. Specifically, § 1002(8) speaks of a “beneficiary” as one “who is or may become entitled to a benefit thereunder.” Indeed, these juxtaposed provisions demonstrate that eligibility and entitlement are distinct terms: § 1002(7) defines a “participant” as one “who is or may become eligible” while § 1002(8) defines a “beneficiary” as one “who is or may become entitled.” Finally, the majority criticizes the dissent for not addressing . But is an inapt case for deciding the specific issue of this case. First, it predates Glenn and Conkright, both of which reinforce the dissent‘s understanding of Firestone. Second, is a preemption case that decided whether a state can “prohibit[] designing an insurance contract so as to accord unfettered discretion to the insurer to interpret the contract‘s terms.” . It held that this type of statute was allowed because, like an insurance contract, the focus was on a legal question—whether a state statute could modify a plan‘s form of legal analysis and not whether the specific person was entitled to money for medical treatment. Third, , as a preemption decision, had nothing to do with enforcement of § 1132(a)(1)(B), the statute at issue here. Consequently, nothing in ‘s holding depended on the language cited by the majority to support its position today. To the point, the case serves neither the majority nor the dissent. See .
10
Restatement (Second) of Trusts § 187 cmt. c (1959).
11
Id. at cmt. c, which provides in its entirety that:

c. Kinds of discretionary powers. The rule stated in this Section is applicable both to the powers of managing the trust estate conferred upon the trustee either in specific words or otherwise, and also to such powers as may be conferred upon him to determine the disposition of the beneficial interest. Thus, it is applicable not only to powers to lease, sell or mortgage the trust property or to invest trust funds, but alsо to powers to allocate the beneficial interest among various beneficiaries, to determine the amount necessary for a beneficiary‘s support, or to terminate the trust.

12
Id. at cmt. d:

d. Factors in determining whether there is an abuse of discretion. In determining the question whether the trustee is guilty of an abuse of discretion in exercising or failing to exercise a power, the following circumstances may be relevant: (1) the extent of the discretion conferred upon the trustee by the terms of the trust; (2) the purposes of the trust; (3) the nature of the power; (4) the existence or non-existence, the definiteness or indefiniteness, of an external standard by which the reasonableness of the trustee‘s conduct can be judged; (5) the motives of the trustee in exercising or refraining from exercising the power; (6) the existence or nonexistence of an interest in the trustee conflicting with that of the beneficiaries.

13
Id. at cmt. e:

e. No abuse of discretion. If discretion is conferred upon the trustee in the exercise of a power, the court will not interfere unless the trustee in exercising or failing to exercise the power acts dishonestly, or with an improper even though not a dishonest motive, or fails to use his judgment, or acts beyond the bounds of a reasonable judgment. The mere fact that if the discretion had been conferred upon the сourt, the court would have exercised the power differently, is not a sufficient reason for interfering with the exercise of the power by the trustee. Thus, if the trustee is empowered to apply so much of the trust property as he may deem necessary for the support of the beneficiary, the court will not interfere with the discretion of the trustee on the ground that he has applied too small an amount, if in the exercise of his judgment honestly and with proper motives he applies at least the minimum amount which could reasonably be considered necessary, even though if the matter were left to the court determine in its discretion it might have applied a larger amount. So also, the court will not interfere on the ground that the trustee has applied too large an amount, if in the exercise of his judgment honestly and with proper motives he applies an amount not greater than a reasonable person might deem necessary for the beneficiary‘s support, although the amount is greater than the court would itself have awarded.

15
(emphasis omitted).
*
Judge Jolly, now a Senior Judge of this court, participated in the consideration of this en banc case. Judges Willett and Ho were not on the court when this case was heard en banc.
Read the detailed case summary