The question in this case is whether treatment decisions made by a health maintenance organization, acting through its physician employees, are fiduciary acts within the meaning of the Employee Retirement Income Security Act of 1974 (ERISA), 88 Stat. 832, as amended, 29 U.S.C. § 1001 et seq. (1994 ed. and Supp. III). We hold that they are not.
I
Petitioners, Carle Clinic Association, P. C., Health Alliance Medical Plans, Inc., and Carle Health Insurance Management Co., Inc. (collectively Carle) function as a health maintenance organization (HMO) organized for profit. Its owners are physicians providing prepaid medical services to participants whose employers contract with Carle to provide such coverage. Respondent, Cynthia Herdrich, was covered by Carle through her husband’s employer, State Farm Insurance Company.
The events in question began when a Carle physician, petitioner Lori Pegram, examined Herdrich, who was experiencing pain in the midline area of her groin. Six days later, Dr. Pegram discovered a six by eight centimeter inflamed mass in Herdrich’s abdomen. Despite the noticeable inflammation, Dr. Pegram did not order an ultrasound diagnostic procedure at a local hospital, but decided that Herdrich would have to wait eight more days for an ultrasound , to be performed at a facility staffed by Carle more than 50 miles away. Before the eight days were over, Herdrich’s appendix ruptured, causing peritonitis. See 154 F.3d 362, 365, n. 1 (CA7 1998).
Herdrich sued Pegram and Carle in state court for medical malpractice, and she later added two counts charging state-law fraud. Carle and Pegram responded that ERISA preempted the new counts, and removed the case to federal court,2 where they then sought summary judgment on the state-law fraud counts. The District Court granted their motion as to the second fraud count but granted Herdrich leave to amend the one remaining. This she did by alleging that provision of medical services under the terms of the Carle HMO organization, rewarding its physician owners for limiting medical care, entailed an inherent or anticipatory breach of an ERISA fiduciary duty, since these terms created an incentive to make decisions in the physicians’ self-interest, rather than the exclusive interests of plan participants.3
Herdrich sought relief under 29 U.S.C. § 1109(a), which provides that
“[a]ny person who is a fiduciary with respect to a plan who breaches any of the responsibilities, obligations, or duties imposed upon fiduciaries by this subchapter shall be personally liable to make good to such plan any losses to the plan resulting from each such breach, and to restore to such plan any profits of such fiduciary which have been made through use of assets of the plan by the fiduciary, and shall be subject to such other equitable or remedial relief as the court may deem appropriate, including removal of such fiduciary.”
When Carle moved to dismiss the ERISA count for failure to state a claim upon which relief could be granted, the District Court granted the motion, accepting the Magistrate Judge’s determination that Carle was not “involved [in these events] as” an ERISA fiduciary. App. to Pet. for Cert. 63a. The original malpractice counts were then tried to a jury, and Herdrich prevailed on both, receiving $35,000 in compensation for her injury. 154 F.3d, at 367. She then appealed the dismissal of the ERISA claim to the Court of Appeals for the Seventh Circuit, which reversed. The court held that Carle was acting as a fiduciary when its physicians made the challenged decisions and that Herdrich’s allegations were sufficient to state a claim:
“Our decision does not stand for the proposition that the existence of incentives automatically gives rise to a breach of fiduciary duty. Rather, we hold that incentives can rise to the level of a breach where, as pleaded here, the fiduciary trust between plan participants and plan fiduciaries no longer exists (i.e., where physicians delay providing necessary treatment to, or withhold administering proper care to, plan beneficiaries for the sole purpose of increasing their bonuses).” Id., at 373.
We granted certiorari, 527 U.S. 1068 (1999), and now reverse the Court of Appeals.
II
Whether Carle is a fiduciary when it acts through its physician owners as pleaded in the ERISA count depends on some background of fact and law about HMO organizations, medical benefit plans, fiduciary obligation, and the meaning of Herdrich’s allegations.
A
Traditionally, medical care in the United States has been provided on a “fee-for-service” basis. A physician charges so much for a general physical exam, a vaccination, a tonsillectomy, and so on. The physician bills the patient for services provided or, if there is insurance and the doctor is willing, submits the bill for the patient’s care to the insurer, for payment subject to the terms of the insurance agreement. Cf. R. Rosenblatt, S. Law, & S. Rosenbaum, Law and the American Health Care System 543—544 (1997) (hereinafter Rosenblatt) (citing Weiner & de Lissovoy, Razing a Tower of Babel: A Taxonomy for Managed Care and Health Insurance Plans, 18 J. Health Politics, Policy & Law 75, 76—78 (Summer 1993)). In a fee-for-service system, a physician’s financial incentive is to provide more care, not less, so long as payment is forthcoming. The check on this incentive is a physician’s obligation to exercise reasonable medical skill and judgment in the patient’s interest.
Beginning in the late 1960’s, insurers and others developed new models for health-care delivery, including HMOs. Cf. Rosenblatt 546. The defining feature of an HMO is receipt of a fixed fee for each patient enrolled under the terms of a contract to provide specified health care if needed. The HMO thus assumes the financial risk of providing the benefits promised: if a participant never gets sick, the HMO keeps the money regardless, and if a participant becomes expensively ill, the HMO is responsible for the treatment agreed upon even if its cost exceeds the participant’s premiums.
Like other risk-bearing organizations, HMOs take steps to control costs. At the least, HMOs, like traditional insurers, will in some fashion make coverage determinations, scrutinizing requested services against the contractual provisions to make sure that a request for care falls within the scope of covered circumstances (pregnancy, for example), or that a given treatment falls within the scope of the care promised (surgery, for instance). They customarily issue general guidelines for their physicians about appropriate levels of care. See id., at 568—570. And they commonly require utilization review (in which specific treatment decisions are reviewed by a decisionmaker other than the treating physician) and approval in advance (precertification) for many types of care, keyed to standards of medical necessity or the reasonableness of the proposed treatment. See Andreson, Is Utilization Review the Practice of Medicine?, Implications for Managed Care Administrators, 19 J. Legal Med. 431, 432 (Sept. 1998). These cost-controlling measures are commonly complemented by specific financial incentives to physicians, rewarding them for decreasing utilization of health-care services, and penalizing them for what may be found to be excessive treatment, see Rosenblatt 563—565; John K. Iglehart, Health Policy Report: The American Health Care System–Managed Care, 327 New England J. Med. 742, 742—747 (1992). Hence, in an HMO system, a physician’s financial interest lies in providing less care, not more. The check on this influence (like that on the converse, fee-for-service incentive) is the professional obligation to provide covered services with a reasonable degree of skill and judgment in the patient’s interest. See Brief for American Medical Association as Amicus Curiae 17—21.
The adequacy of professional obligation to counter
financial self-interest has been challenged no matter what the form of medical
organization. HMOs became popular because fee-for-service physicians were
thought to be providing unnecessary or useless services; today, many doctors and
other observers argue that HMOs often ignore the individual needs of a patient
in order to improve the HMOs’ bottom lines. See, e.
B
Herdrich focuses on the Carle scheme’s provision for a “year-end distribution,” n. 3, supra, to the HMO’s physician owners. She argues that this particular incentive device of annually paying physician owners the profit resulting from their own decisions rationing care can distinguish Carle’s organization from HMOs generally, so that reviewing Carle’s decisions under a fiduciary standard as pleaded in Herdrich’s complaint would not open the door to like claims about other HMO structures. While the Court of Appeals agreed, we think otherwise, under the law as now written.
Although it is true that the relationship between sparing medical treatment and physician reward is not a subtle one under the Carle scheme, no HMO organization could survive without some incentive connecting physician reward with treatment rationing. The essence of an HMO is that salaries and profits are limited by the HMO’s fixed membership fees. See Orentlicher, Paying Physicians More To Do Less: Financial Incentives to Limit Care, 30 U. Rich. L. Rev. 155, 174 (1996). This is not to suggest that the Carle provisions are as socially desirable as some other HMO organizational schemes; they may not be. See, e.g., Grumbach, Osmond, Vranigan, Jaffe, & Bindman, Primary Care Physicians’ Experience of Financial Incentives in Managed-Care Systems, 339 New Eng. J. Med. 1516 (1998) (arguing that HMOs that reward quality of care and patient satisfaction would be preferable to HMOs that reward only physician productivity). But whatever the HMO, there must be rationing and inducement to ration.
Since inducement to ration care goes to the very point of any HMO scheme, and rationing necessarily raises some risks while reducing others (ruptured appendixes are more likely; unnecessary appendectomies are less so), any legal principle purporting to draw a line between good and bad HMOs would embody, in effect, a judgment about socially acceptable medical risk. A valid conclusion of this sort would, however, necessarily turn on facts to which courts would probably not have ready access: correlations between malpractice rates and various HMO models, similar correlations involving fee-for-service models, and so on. And, of course, assuming such material could be obtained by courts in litigation like this, any standard defining the unacceptably risky HMO structure (and consequent vulnerability to claims like Herdrich’s) would depend on a judgment about the appropriate level of expenditure for health care in light of the associated malpractice risk. But such complicated factfinding and such a debatable social judgment are not wisely required of courts unless for some reason resort cannot be had to the legislative process, with its preferable forum for comprehensive investigations and judgments of social value, such as optimum treatment levels and health care expenditure. Cf. Turner Broadcasting System, Inc. v. FCC, 512 U.S. 622, 665—666 (1994) (opinion of Kennedy, J.) (“Congress is far better equipped than the judiciary to ‘amass and evaluate the vast amounts of data’ bearing upon an issue as complex and dynamic as that presented here” (quoting Walters v. National Assn. of Radiation Survivors, 473 U.S. 305, 331, n. 12 (1985))); Patsy v. Board of Regents of Fla., 457 U.S. 496, 513 (1982) (“[T]he relevant policy considerations do not invariably point in one direction, and there is vehement disagreement over the validity of the assumptions underlying many of them. The very difficulty of these policy considerations, and Congress’ superior institutional competence to pursue this debate, suggest that legislative not judicial solutions are preferable” (footnote omitted)).
We think, then, that courts are not in a position to derive a sound legal principle to differentiate an HMO like Carle from other HMOs.5 For that reason, we proceed on the assumption that the decisions listed in Herdrich’s complaint cannot be subject to a claim that they violate fiduciary standards unless all such decisions by all HMOs acting through their owner or employee physicians are to be judged by the same standards and subject to the same claims.
C
We turn now from the structure of HMOs to the requirements of ERISA. A fiduciary within the meaning of ERISA must be someone acting in the capacity of manager, administrator, or financial adviser to a “plan,” see 29 U.S.C. § 1002(21)(A)(i)—(iii), and Herdich’s ERISA count accordingly charged Carle with a breach of fiduciary duty in discharging its obligations under State Farm’s medical plan. App. to Pet. for Cert. 85a—86a. ERISA’s definition of an employee welfare benefit plan is ultimately circular: “any plan, fund, or program … to the extent that such plan, fund, or program was established … for the purpose of providing … through the purchase of insurance or otherwise … medical, surgical, or hospital care or benefits.” §1002(1)(A). One is thus left to the common understanding of the word “plan” as referring to a scheme decided upon in advance, see Webster’s New International Dictionary 1879 (2d ed. 1957); Jacobson & Pomfret, Form, Function, and Managed Care Torts: Achieving Fairness and Equity in ERISA Jurisprudence, 35 Houston L. Rev. 985, 1050 (1998). Here the scheme comprises a set of rules that define the rights of a beneficiary and provide for their enforcement. Rules governing collection of premiums, definition of benefits, submission of claims, and resolution of disagreements over entitlement to services are the sorts of provisions that constitute a plan. See Hansen v. Continental Ins. Co., 940 F.2d 971, 974 (CA5 1991).
Thus, when employers contract with an HMO to provide benefits to employees subject to ERISA, the provisions of documents that set up the HMO are not, as such, an ERISA plan, but the agreement between an HMO and an employer who pays the premiums may, as here, provide elements of a plan by setting out rules under which beneficiaries will be entitled to care.
D
As just noted, fiduciary obligations can apply to managing, advising, and administering an ERISA plan, the fiduciary function addressed by Herdrich’s ERISA count being the exercise of “discretionary authority or discretionary responsibility in the administration of [an ERISA] plan,” 29 U.S.C. § 1002(21)(A)(iii). And as we have already suggested, although Carle is not an ERISA fiduciary merely because it administers or exercises discretionary authority over its own HMO business, it may still be a fiduciary if it administers the plan.
In general terms, fiduciary responsibility under ERISA is simply stated. The statute provides that fiduciaries shall discharge their duties with respect to a plan “solely in the interest of the participants and beneficiaries,” §1104(a)(1), that is, “for the exclusive purpose of (i) providing benefits to participants and their beneficiaries; and (ii) defraying reasonable expenses of administering the plan,” §1104(a)(1)(A). 6 These responsibilities imposed by ERISA have the familiar ring of their source in the common law of trusts. See Central States, Southeast & Southwest Areas Pension Fund v. Central Transport, Inc., 472 U.S. 559, 570 (1985) (“[R]ather than explicitly enumerating all of the powers and duties of trustees and other fiduciaries, Congress invoked the common law of trusts to define the general scope of their authority and responsibility”). Thus, the common law (understood as including what were once the distinct rules of equity) charges fiduciaries with a duty of loyalty to guarantee beneficiaries’ interests: “The most fundamental duty owed by the trustee to the beneficiaries of the trust is the duty of loyalty… . It is the duty of a trustee to administer the trust solely in the interest of the beneficiaries.” 2A A. Scott & W. Fratcher, Trusts §170, 311 (4th ed. 1987) (hereinafter Scott); see also G. Bogert & G. Bogert, Law of Trusts and Trustees §543 (rev. 2d ed. 1980) (“Perhaps the most fundamental duty of a trustee is that he must display throughout the administration of the trust complete loyalty to the interests of the beneficiary and must exclude all selfish interest and all consideration of the interests of third persons”); Central States, supra, at 570—571; Meinhard v. Salmon, 249 N. Y. 458, 464, 164 N. E. 545, 546 (1928) (Cardozo, J.) (“Many forms of conduct permissible in a workaday world for those acting at arm’s length, are forbidden to those bound by fiduciary ties. A trustee is held to something stricter than the morals of the market place. Not honesty alone, but the punctilio of an honor the most sensitive, is then the standard of behavior”).
Beyond the threshold statement of responsibility, however, the analogy between ERISA fiduciary and common law trustee becomes problematic. This is so because the trustee at common law characteristically wears only his fiduciary hat when he takes action to affect a beneficiary, whereas the trustee under ERISA may wear different hats.
Speaking of the traditional trustee, Professor Scott’s treatise admonishes that the trustee “is not permitted to place himself in a position where it would be for his own benefit to violate his duty to the beneficiaries.” 2A Scott, §170, at 311. Under ERISA, however, a fiduciary may have financial interests adverse to beneficiaries. Employers, for example, can be ERISA fiduciaries and still take actions to the disadvantage of employee beneficiaries, when they act as employers (e.g., firing a beneficiary for reasons unrelated to the ERISA plan), or even as plan sponsors (e.g., modifying the terms of a plan as allowed by ERISA to provide less generous benefits). Nor is there any apparent reason in the ERISA provisions to conclude, as Herdrich argues, that this tension is permissible only for the employer or plan sponsor, to the exclusion of persons who provide services to an ERISA plan.
ERISA does require, however, that the fiduciary with two hats wear only one at a time, and wear the fiduciary hat when making fiduciary decisions. See Hughes Aircraft Co. v. Jacobson, 525 U.S. 432, 443—444 (1999); Varity Corp. v. Howe, 516 U.S. 489, 497 (1996). Thus, the statute does not describe fiduciaries simply as administrators of the plan, or managers or advisers. Instead it defines an administrator, for example, as a fiduciary only “to the extent” that he acts in such a capacity in relation to a plan. 29 U.S.C. § 1002(21)(A). In every case charging breach of ERISA fiduciary duty, then, the threshold question is not whether the actions of some person employed to provide services under a plan adversely affected a plan beneficiary’s interest, but whether that person was acting as a fiduciary (that is, was performing a fiduciary function) when taking the action subject to complaint.
E
The allegations of Herdrich’s ERISA count that identify the claimed fiduciary breach are difficult to understand. In this count, Herdrich does not point to a particular act by any Carle physician owner as a breach. She does not complain about Pegram’s actions, and at oral argument her counsel confirmed that the ERISA count could have been brought, and would have been no different, if Herdrich had never had a sick day in her life. Tr. of Oral Arg. 53—54.
What she does claim is that Carle, acting through its physician owners, breached its duty to act solely in the interest of beneficiaries by making decisions affecting medical treatment while influenced by the terms of the Carle HMO scheme, under which the physician owners ultimately profit from their own choices to minimize the medical services provided. She emphasizes the threat to fiduciary responsibility in the Carle scheme’s feature of a year-end distribution to the physicians of profit derived from the spread between subscription income and expenses of care and administration. App. to Pet. for Cert. 86a.
The specific payout detail of the plan was, of course, a feature that the employer as plan sponsor was free to adopt without breach of any fiduciary duty under ERISA, since an employer’s decisions about the content of a plan are not themselves fiduciary acts. Lockheed Corp. v. Spink, 517 U.S. 882, 887 (1996) (“Nothing in ERISA requires employers to establish employee benefit plans. Nor does ERISA mandate what kind of benefit employers must provide if they choose to have such a plan”).7 Likewise it is clear that there was no violation of ERISA when the incorporators of the Carle HMO provided for the year-end payout. The HMO is not the ERISA plan, and the incorporation of the HMO preceded its contract with the State Farm plan. See 29 U.S.C. § 1109(b) (no fiduciary liability for acts preceding fiduciary status).
The nub of the claim, then, is that when State Farm contracted with Carle, Carle became a fiduciary under the plan, acting through its physicians. At once, Carle as fiduciary administrator was subject to such influence from the year-end payout provision that its fiduciary capacity was necessarily compromised, and its readiness to act amounted to anticipatory breach of fiduciary obligation.
F
The pleadings must also be parsed very carefully to understand what acts by physician owners acting on Carle’s behalf are alleged to be fiduciary in nature.8 It will help to keep two sorts of arguably administrative acts in mind. Cf. Dukes v. U.S. Healthcare, Inc., 57 F.3d 350, 361 (CA3 1995) (discussing dual medical/administrative roles of HMOs). What we will call pure “eligibility decisions” turn on the plan’s coverage of a particular condition or medical procedure for its treatment. “Treatment decisions,” by contrast, are choices about how to go about diagnosing and treating a patent’s condition: given a patient’s constellation of symptoms, what is the appropriate medical response?
These decisions are often practically inextricable from one another, as amici on both sides agree. See Brief for Washington Legal Foundation as Amicus Curiae 12; Brief of Health Law, Policy, and Ethics Scholars as Amici Curiae 10. This is so not merely because, under a scheme like Carle’s, treatment and eligibility decisions are made by the same person, the treating physician. It is so because a great many and possibly most coverage questions are not simple yes-or-no questions, like whether appendicitis is a covered condition (when there is no dispute that a patient has appendicitis), or whether acupuncture is a covered procedure for pain relief (when the claim of pain is unchallenged). The more common coverage question is a when-and-how question. Although coverage for many conditions will be clear and various treatment options will be indisputably compensable, physicians still must decide what to do in particular cases. The issue may be, say, whether one treatment option is so superior to another under the circumstances, and needed so promptly, that a decision to proceed with it would meet the medical necessity requirement that conditions the HMO’s obligation to provide or pay for that particular procedure at that time in that case. The Government in its brief alludes to a similar example when it discusses an HMO’s refusal to pay for emergency care on the ground that the situation giving rise to the need for care was not an emergency, Brief for United States as Amicus Curiae 20—21.9 In practical terms, these eligibility decisions cannot be untangled from physicians’ judgments about reasonable medical treatment, and in the case before us, Dr. Pegram’s decision was one of that sort. She decided (wrongly, as it turned out) that Herdrich’s condition did not warrant immediate action; the consequence of that medical determination was that Carle would not cover immediate care, whereas it would have done so if Dr. Pegram had made the proper diagnosis and judgment to treat. The eligibility decision and the treatment decision were inextricably mixed, as they are in countless medical administrative decisions every day.
The kinds of decisions mentioned in Herdrich’s ERISA count and claimed to be fiduciary in character are just such mixed eligibility and treatment decisions: physicians’ conclusions about when to use diagnostic tests; about seeking consultations and making referrals to physicians and facilities other than Carle’s; about proper standards of care, the experimental character of a proposed course of treatment, the reasonableness of a certain treatment, and the emergency character of a medical condition.
We do not read the ERISA count, however, as alleging fiduciary breach with reference to a different variety of administrative decisions, those we have called pure eligibility determinations, such as whether a plan covers an undisputed case of appendicitis. Nor do we read it as claiming breach by reference to discrete administrative decisions separate from medical judgments; say, rejecting a claim for no other reason than the HMO’s financial condition. The closest Herdrich’s ERISA count comes to stating a claim for a pure, unmixed eligibility decision is her general allegation that Carle determines “which claims are covered under the Plan and to what extent,” App. to Pet. for Cert. 86a. But this vague statement, difficult to interpret in isolation, is given content by the other elements of the complaint, all of which refer to decisions thoroughly mixed with medical judgment. Cf. 5A C. Wright & A. Miller, Federal Practice and Procedure §1357, pp. 320—321 (1990) (noting that, where specific allegations clarify the meaning of broader allegations, they may be used to interpret the complaint as a whole). Any lingering uncertainty about what Herdrich has in mind is dispelled by her brief, which explains that this allegation, like the others, targets medical necessity determinations. Brief for Respondent 19; see also id., at 3. 10
III
A
Based on our understanding of the matters just discussed, we think Congress did not intend Carle or any other HMO to be treated as a fiduciary to the extent that it makes mixed eligibility decisions acting through its physicians. We begin with doubt that Congress would ever have thought of a mixed eligibility decision as fiduciary in nature. At common law, fiduciary duties characteristically attach to decisions about managing assets and distributing property to beneficiaries. See Bogert & Bogert, Law of Trusts and Trustees, §§551, 741—747, 751—775, 781—799; 2A Scott, §§176, 181, 3 id., §§188—193, 3A id., §232. Trustees buy, sell, and lease investment property, lend and borrow, and do other things to conserve and nurture assets. They pay out income, choose beneficiaries, and distribute remainders at termination. Thus, the common law trustee’s most defining concern historically has been the payment of money in the interest of the beneficiary.
Mixed eligibility decisions by an HMO acting through its physicians have, however, only a limited resemblance to the usual business of traditional trustees. To be sure, the physicians (like regular trustees) draw on resources held for others and make decisions to distribute them in accordance with entitlements expressed in a written instrument (embodying the terms of an ERISA plan). It is also true that the objects of many traditional private and public trusts are ultimately the same as the ERISA plans that contract with HMOs. Private trusts provide medical care to the poor; thousands of independent hospitals are privately held and publicly accountable trusts, and charitable foundations make grants to stimulate the provision of health services. But beyond this point the resemblance rapidly wanes. Traditional trustees administer a medical trust by paying out money to buy medical care, whereas physicians making mixed eligibility decisions consume the money as well. Private trustees do not make treatment judgments, whereas treatment judgments are what physicians reaching mixed decisions do make, by definition. Indeed, the physicians through whom HMOs act make just the sorts of decisions made by licensed medical practitioners millions of times every day, in every possible medical setting: HMOs, fee-for-service proprietorships, public and private hospitals, military field hospitals, and so on. The settings bear no more resemblance to trust departments than a decision to operate turns on the factors controlling the amount of a quarterly income distribution. Thus, it is at least questionable whether Congress would have had mixed eligibility decisions in mind when it provided that decisions administering a plan were fiduciary in nature. Indeed, when Congress took up the subject of fiduciary responsibility under ERISA, it concentrated on fiduciaries’ financial decisions, focusing on pension plans, the difficulty many retirees faced in getting the payments they expected, and the financial mismanagement that had too often deprived employees of their benefits. See, e.g., S. Rep. No. 93—127, p. 5 (1973); S. Rep. No. 93—383, p. 17 (1973); id., at 95. Its focus was far from the subject of Herdrich’s claim.
Our doubt that Congress intended the category of fiduciary administrative functions to encompass the mixed determinations at issue here hardens into conviction when we consider the consequences that would follow from Herdrich’s contrary view.
B
First, we need to ask how this fiduciary standard would affect HMOs if it applied as Herdrich claims it should be applied, not directed against any particular mixed decision that injured a patient, but against HMOs that make mixed decisions in the course of providing medical care for profit. Recovery would be warranted simply upon showing that the profit incentive to ration care would generally affect mixed decisions, in derogation of the fiduciary standard to act solely in the interest of the patient without possibility of conflict. Although Herdrich is vague about the mechanics of relief, the one point that seems clear is that she seeks the return of profit from the pockets of the Carle HMO’s owners, with the money to be given to the plan for the benefit of the participants. See 29 U.S.C. § 1109(a) (return of all profits is an appropriate ERISA remedy). Since the provision for profit is what makes the HMO a proprietary organization, her remedy in effect would be nothing less than elimination of the for-profit HMO. Her remedy might entail even more than that, although we are in no position to tell whether and to what extent nonprofit HMO schemes would ultimately survive the recognition of Herdrich’s theory.11 It is enough to recognize that the Judiciary has no warrant to precipitate the upheaval that would follow a refusal to dismiss Herdrich’s ERISA claim. The fact is that for over 27 years the Congress of the United States has promoted the formation of HMO practices. The Health Maintenance Organization Act of 1973, 87 Stat. 914, 42 U.S.C. § 300e et seq., allowed the formation of HMOs that assume financial risks for the provision of health care services, and Congress has amended the Act several times, most recently in 1996. See 110 Stat. 1976, codified at 42 U.S.C. § 300e (1994 ed, Supp. III). If Congress wishes to restrict its approval of HMO practice to certain preferred forms, it may choose to do so. But the Federal Judiciary would be acting contrary to the congressional policy of allowing HMO organizations if it were to entertain an ERISA fiduciary claim portending wholesale attacks on existing HMOs solely because of their structure, untethered to claims of concrete harm.
C
The Court of Appeals did not purport to entertain quite the broadside attack that Herdrich’s ERISA claim thus entails, see 154 F.3d, at 373, and the second possible consequence of applying the fiduciary standard that requires our attention would flow from the difficulty of extending it to particular mixed decisions that on Herdrich’s theory are fiduciary in nature.
The fiduciary is, of course, obliged to act exclusively in the interest of the beneficiary, but this translates into no rule readily applicable to HMO decisions or those of any other variety of medical practice. While the incentive of the HMO physician is to give treatment sparingly, imposing a fiduciary obligation upon him would not lead to a simple default rule, say, that whenever it is reasonably possible to disagree about treatment options, the physician should treat aggressively. After all, HMOs came into being because some groups of physicians consistently provided more aggressive treatment than others in similar circumstances, with results not perceived as justified by the marginal expense and risk associated with intervention; excessive surgery is not in the patient’s best interest, whether provided by fee-for-service surgeons or HMO surgeons subject to a default rule urging them to operate. Nor would it be possible to translate fiduciary duty into a standard that would allow recovery from an HMO whenever a mixed decision influenced by the HMO’s financial incentive resulted in a bad outcome for the patient. It would be so easy to allege, and to find, an economic influence when sparing care did not lead to a well patient, that any such standard in practice would allow a factfinder to convert an HMO into a guarantor of recovery.
These difficulties may have led the Court of Appeals to try to confine the fiduciary breach to cases where “the sole purpose” of delaying or withholding treatment was to increase the physician’s financial reward, ibid. But this attempt to confine mixed decision claims to their most egregious examples entails erroneous corruption of fiduciary obligation and would simply lead to further difficulties that we think fatal. While a mixed decision made solely to benefit the HMO or its physician would violate a fiduciary duty, the fiduciary standard condemns far more than that, in its requirement of “an eye single” toward beneficiaries’ interests, Donovan v. Bierwirth, 680 F.2d 263, 271 (CA2 1982). But whether under the Court of Appeals’s rule or a straight standard of undivided loyalty, the defense of any HMO would be that its physician did not act out of financial interest but for good medical reasons, the plausibility of which would require reference to standards of reasonable and customary medical practice in like circumstances. That, of course, is the traditional standard of the common law. See W. Keeton, D. Dobbs, R. Keeton, & D. Owens, Prosser and Keeton on Law of Torts §32, pp. 188—189 (5th ed. 1984). Thus, for all practical purposes, every claim of fiduciary breach by an HMO physician making a mixed decision would boil down to a malpractice claim, and the fiduciary standard would be nothing but the malpractice standard traditionally applied in actions against physicians.
What would be the value to the plan participant of having this kind of ERISA fiduciary action? It would simply apply the law already available in state courts and federal diversity actions today, and the formulaic addition of an allegation of financial incentive would do nothing but bring the same claim into a federal court under federal-question jurisdiction. It is true that in States that do not allow malpractice actions against HMOs the fiduciary claim would offer a plaintiff a further defendant to be sued for direct liability, and in some cases the HMO might have a deeper pocket than the physician. But we have seen enough to know that ERISA was not enacted out of concern that physicians were too poor to be sued, or in order to federalize malpractice litigation in the name of fiduciary duty for any other reason. It is difficult, in fact, to find any advantage to participants across the board, except that allowing them to bring malpractice actions in the guise of federal fiduciary breach claims against HMOs would make them eligible for awards of attorney’s fees if they won. See 29 U.S.C. § 1132(g)(1). But, again, we can be fairly sure that Congress did not create fiduciary obligations out of concern that state plaintiffs were not suing often enough, or were paying too much in legal fees.
The mischief of Herdrich’s position would, indeed, go further than mere replication of state malpractice actions with HMO defendants. For not only would an HMO be liable as a fiduciary in the first instance for its own breach of fiduciary duty committed through the acts of its physician employee, but the physician employee would also be subject to liability as a fiduciary on the same basic analysis that would charge the HMO. The physician who made the mixed administrative decision would be exercising authority in the way described by ERISA and would therefore be deemed to be a fiduciary. See 29 CFR §§2509.75—5, Question D1; 2509.75—8, Question D-3 (1993) (stating that an individual who exercises authority on behalf of an ERISA fiduciary in interpreting and administering a plan will be deemed a fiduciary). Hence the physician, too, would be subject to suit in federal court applying an ERISA standard of reasonable medical skill. This result, in turn, would raise a puzzling issue of preemption. On its face, federal fiduciary law applying a malpractice standard would seem to be a prescription for preemption of state malpractice law, since the new ERISA cause of action would cover the subject of a state-law malpractice claim. See 29 U.S.C. § 1144 (preempting state laws that “relate to [an] employee benefit plan”). To be sure, New York State Conference of Blue Cross & Blue Shield Plans v. Travelers Ins. Co., 514 U.S. 645, 654—655 (1995), throws some cold water on the preemption theory; there, we held that, in the field of health care, a subject of traditional state regulation, there is no ERISA preemption without clear manifestation of congressional purpose. But in that case the convergence of state and federal law was not so clear as in the situation we are positing; the state-law standard had not been subsumed by the standard to be applied under ERISA. We could struggle with this problem, but first it is well to ask, again, what would be gained by opening the federal courthouse doors for a fiduciary malpractice claim, save for possibly random fortuities such as more favorable scheduling, or the ancillary opportunity to seek attorney’s fees. And again, we know that Congress had no such haphazard boons in prospect when it defined the ERISA fiduciary, nor such a risk to the efficiency of federal courts as a new fiduciary-malpractice jurisdiction would pose in welcoming such unheard-of fiduciary litigation.
IV
We hold that mixed eligibility decisions by HMO physicians are not fiduciary decisions under ERISA. Herdrich’s ERISA count fails to state an ERISA claim, and the judgment of the Court of Appeals is reversed.
It is so ordered.
1. Although Lori Pegram, a physician owner of Carle, is listed as a petitioner, it is unclear to us that she retains a direct interest in the outcome of this case.
2. Herdrich does not contest the propriety of removal before us, and we take no position on whether or not the case was properly removed. As we will explain, Herdrich’s amended complaint alleged ERISA violations, over which the federal courts have jurisdiction, and we therefore have jurisdiction regardless of the correctness of the removal. See Grubbs v. General Elec. Credit Corp., 405 U.S. 699 (1972); Mackay v. Uinta Development Co., 229 U.S. 173 (1913).
3. The specific allegations were these: “11. Defendants are fiduciaries with respect to the Plan and under 29 [U.S.C. §]1109(a) are obligated to discharge their duties with respect to the Plan solely in the interest of the participants and beneficiaries and “a. for the exclusive purpose of: “i. providing benefits to participants and their beneficiaries; and “ii. defraying reasonable expenses of administering the Plan; “b. with the care, skill, prudence, and diligence under the circumstances then prevailing that a prudent man acting in a like capacity and familiar with such matters would use in the conduct of an enterprise of a like character and like aims. “12. In breach of that duty: “a. CARLE owner/physicians are the officers and directors of HAMP and CHIMCO and receive a year-end distribution, based in large part upon, supplemental medical expense payments made to CARLE by HAMP and CHIMCO; “b. Both HAMP and CHIMCO are directed and controlled by CARLE owner/physicians and seek to fund their supplemental medical expense payments to CARLE: “i. by contracting with CARLE owner/physicians to provide the medical services contemplated in the Plan and then having those contracted owner/physicians: “(1) minimize the use of diagnostic tests; “(2) minimize the use of facilities not owned by CARLE; and “(3) minimize the use of emergency and non-emergency consultation and/or referrals to non-contracted physicians. “ii. by administering disputed and non-routine health insurance claims and determining: “(1) which claims are covered under the Plan and to what extent; “(2) what the applicable standard of care is; “(3) whether a course of treatment is experimental; “(4) whether a course of treatment is reasonable and customary; and “(5) whether a medical condition is an emergency.” App to Pet. for Cert. 85a—86a.
4. There are, of course, contrary perspectives, and we endorse neither side of the debate today.
5. They are certainly not capable of making that distinction on a motion to dismiss; if we accepted the Court of Appeals’s reasoning, complaints against any flavor of HMO would have to proceed at least to the summary judgment stage.
6. In addition, fiduciaries must discharge their duties “(B) with the care, skill, prudence, and diligence under the circumstances then prevailing that a prudent man acting in a like capacity and familiar with such matters would use in the conduct of an enterprise of a like character and with like aims; “(C) by diversifying the investments of the plan so as to minimize the risk of large losses, unless under the circumstances it is clearly prudent not to do so; and “(D) in accordance with the documents and instruments governing the plan insofar as such documents and instruments are consistent with the provisions of this subchapter and subchapter III of this chapter.” 29 U.S.C. § 1104(a)(1).
7. It does not follow that those who administer a particular plan design may not have difficulty in following fiduciary standards if the design is awkward enough. A plan might lawfully provide for a bonus for administrators who denied benefits to every 10th beneficiary, but it would be difficult for an administrator who received the bonus to defend against the claim that he had not been solely attentive to the beneficiaries’ interests in carrying out his administrative duties. The important point is that Herdrich is not suing the employer, State Farm, and her claim cannot be analyzed as if she were.
8. Herdrich argues that Carle is judicially estopped from denying its fiduciary status as to the relevant decisions, because it sought and sucessfully defended removal of Herdrich’s state action to the Federal District Court on the ground that it was a fiduciary with respect to Herdrich’s fraud claims. Judicial estoppel generally prevents a party from prevailing in one phase of a case on an argument and then relying on a contradictory argument to prevail in another phase. See Rissetto v. Plumbers & Steamfitters Local 343, 94 F.3d 597, 605 (CA9 1996). The fraud claims in Herdrich’s initial complaint, however, could be read to allege breach of a fiduciary obligation to disclose physician incentives to limit care, whereas her amended complaint alleges an obligation to avoid such incentives. Although we are not presented with the issue here, it could be argued that Carle is a fiduciary insofar as it has discretionary authority to administer the plan, and so it is obligated to disclose characteristics of the plan and of those who provide services to the plan, if that information affects beneficiaries’ material interests. See, e.g., Glaziers and Glassworkers Union Local No. 252 Annuity Fund v. Newbridge Securities, Inc., 93 F.3d 1171, 1179—1181 (CA3 1996) (discussing the disclosure obligations of an ERISA fiduciary); cf. Varity Corp. v. Howe, 516 U.S. 489, 505 (1996) (holding that ERISA fiduciaries may have duties to disclose information about plan prospects that they have no duty, or even power, to change). But failure to disclose is no longer the allegation of the amended complaint. Because fiduciary duty to disclose is not necessarily coextensive with fiduciary responsibility for the subject matter of the disclosure, Carle is not estopped from contesting its fiduciary status with respect to the allegations of the amended complaint.
9. ERISA makes separate provision for suits to receive particular benefits. See 29 U.S.C. § 1132(a)(1)(B). We have no occasion to discuss the standards governing such a claim by a patient who, as in the example in text, was denied reimbursement for emergency care. Nor have we reason to discuss the interaction of such a claim with state law causes of action, see infra, at 24—25.
10. Though this case involves a motion to dismiss under Federal Rule of Civil Procedure 12(b)(6), and the complaint should therefore be construed generously, we may use Herdrich’s brief to clarify allegations in her complaint whose meaning is unclear. See C. Wright & A. Miller, Federal Practice and Procedure, §1364, pp. 480—481 (1990); Southern Cross Overseas Agencies, Inc. v. Wah Kwong Shipping Group Ltd., 181 F.3d 410, 428, n. 8 (CA3 1999); Alicke v. MCI Communications Corp., 111 F.3d 909, 911 (CADC 1997); Early v. Bankers Life & Cas. Co., 959 F.2d 75, 79 (CA7 1992).
11. Herdrich’s theory might well portend the end of nonprofit HMOs as well, since those HMOs can set doctors’ salaries. A claim against a nonprofit HMO could easily allege that salaries were excessively high because they were funded by limiting care, and some nonprofits actually use incentive schemes similar to that challenged here, see Pulvers v. Kaiser Foundation Health Plan, 99 Cal. App. 3d 560, 565, 160 Cal. Rptr. 392, 393—394 (1979) (rejecting claim against nonprofit HMO based on physician incentives). See Brody, Agents Without Principals: The Economic Convergence of the Nonprofit and For-Profit Organizational Forms, 40 N. Y. L. S. L. Rev. 457, 493, and n. 152 (1996) (discussing ways in which nonprofit health providers may reward physician employees).