The case is before the court on a petition for rehearing and suggestion for rehearing en banc filed by the defendants-appellees. On consideration of the petition for rehearing and suggestion for rehearing en banc, a vote of the active members of the court was requested. A majority did not favor rehearing en banc. Chief Judge Posner and Circuit Judges Flaum, Easterbrook and Diane P. Wood voted to grant rehearing en banc. A majority of the judges on the original panel voted to deny rehearing en banc.

Accordingly, IT IS ORDERED that the aforesaid petition for rehearing be, and the same is hereby, DENIED.

EASTERBROOK, Circuit Judge, with whom POSNER, Chief Judge, and FLAUM and DIANE P. WOOD, Circuit Judges, join, dissenting from the denial of rehearing en banc.

Physicians employed by Carle Clinic Association, a health maintenance organization (HMO), failed to diagnose Cynthia Herdrich's appendicitis before her appendix ruptured. Peritonitis ensued, and Herdrich has recovered $35,000 in damages for medical malpractice. She wants more, contending that Carle is a "fiduciary" under ERISA because her husband's employer State Farm Insurance Companies provided Carle's plan as a fringe benefit (making it a "welfare benefit plan" under ERISA), and that the divided loyalties at the core of an HMO structure are forbidden by ERISA. Like other HMO systems, Carle collects in advance for a period of care. The less medical services cost, the more an HMO's owners (here, Carle's physicians) have left as profit at the end of the period. According to the panel, this violates 29 U.S.C. § 1104(a)(1)(A).

Like any business, Carle seeks to hold down its costs. Like most other HMOs, Carle does this through devices that have come to be called "managed care." For example, subscribers must receive their medical care from Carle's own physicians if that is at all possible. Herdrich contends that this rule is responsible for her peritonitis: after finding an inflamed mass in her abdomen, a Carle physician scheduled her for an ultrasound examination eight days later at a Carle facility in Urbana, Illinois, rather than arranging for a local hospital in Bloomington to perform that examination immediately. That delay, the jury found in the malpractice case, led to the peritonitis.

When participants in an HMO plan sought to apply the "fraud" label to the money-saving incentive that characterizes the HMO form of organization, we replied that the details of HMO incentives need not be specifically explained to participants in ERISA plans. Anderson v. Humana, Inc., 24 F.3d 889 (7th Cir.1994). The HMO structure differs substantially from traditional fee-for-service medicine in giving the HMO an incentive to skimp on care once an illness is discovered. It is equally true that the HMO system creates an inducement to keep the subscribers healthy as long as possible. An HMO makes its profit from healthy subscribers and thus provides ample preventive and diagnostic care, while many fee-for-service physicians make their living from sick or injured persons. If the HMO creates an incentive to provide too little care once a subscriber becomes seriously ill, the fee-for-service system coupled with insurance provides an incentive to furnish excessive care, for third parties foot the bill. A choice between prepaid and fee-for-service systems is accordingly difficult to make in principle.

What I find troubling about the panel opinion, and why I believe this case should be reheard en banc, is that the panel has condemned HMO and managed-care systems on medical grounds, 154 F.3d 362, 373-79 (7th Cir.1998), and used its view of good medical practice as the basis of a conclusion that the HMO structure violates ERISA. According to the panel, market forces do not constrain the pernicious incentives that HMOs adopt, and it is accordingly necessary to throw the weight of the law behind traditional fee-for-service medicine. Id. at 374-75 ("market forces are insufficient to cure the deleterious affects [sic] of managed care on the health care industry"). This aspect of the panel's approach is almost a 180? turn from what we wrote in Anderson, a case the panel did not cite:

A health maintenance organization (HMO) offers, for a fixed fee, as much medical care as the patient needs. Providers using traditional fee-for-service methods, by contrast, charge for each procedure. Each method creates an unfortunate incentive: a physician receiving a fee for each service has an incentive to run up the bill by furnishing unnecessary care, and an HMO has an incentive to skimp on care (once patients have signed up and paid) in order to save costs. Each incentive encounters countervailing forces: patients, or insurers on their behalf, resist paying the bills for unnecessary services, and HMOs must afford adequate care if they are to attract patients. HMOs also have a reason to deliver excellent preventive medicine. Prevention may reduce the need for costly services later. Competition among the many providers of health care, and between the principal methods of charging for that care, affords additional protection to consumers.

24 F.3d at 890. But let this pass, and suppose that HMOs and other managed-care systems are inferior to available alternatives. Why does ERISA authorize a court to prescribe its view of the best system?

The answer, according to the panel, is that ERISA requires plan administrators to act as fiduciaries, while the HMO structure puts physicians at (financial) odds with their patients. HMOs are of course not unique in this regard; insurers likewise seek to minimize their outlays for medical care and employ managed-care devices to promote thrift. But I am willing to suppose that Carle did not act as Herdrich's "fiduciary" would have. It did not have to.

Under ERISA,

a person is a fiduciary with respect to a plan to the extent (i) he exercises any discretionary authority or discretionary control respecting management of such plan or exercises any authority or control respecting management or disposition of its assets, (ii) he renders investment advice for a fee or other compensation, direct or indirect, with respect to any moneys or other property of such plan, or has any authority or responsibility to do so, or (iii) he has any discretionary authority or discretionary responsibility in the administration of such plan.

29 U.S.C. § 1002(21)(A) (emphasis added). Carle does not manage the State Farm plan or control its assets, so the panel emphasized sub-(iii), concluding that Carle has "discretionary authority or discretionary responsibility in the administration of such plan." See 154 F.3d at 369-71. Discretionary authority is obvious; but does Carle exercise discretion "in the administration of [the] plan", or only in the provision of medical services? This is a fundamental divide, for fiduciary status under ERISA is not an all-or-none affair. A person is a fiduciary only "to the extent" that he does one of the listed things; many major exercises of discretion, such as selecting the plan's terms, are outside of ERISA's fiduciary duties, even though the same person is a fiduciary when implementing the plan. See Hughes Aircraft Co. v. Jacobson, --- U.S. ----, 119 S. Ct. 755, 142 L. Ed. 2d 881 (1999); Lockheed Corp. v. Spink, 517 U.S. 882, 116 S. Ct. 1783, 135 L. Ed. 2d 153 (1996); McNab v. General Motors Corp., 162 F.3d 959 (7th Cir.1998); Johnson v. Georgia-Pacific Corp., 19 F.3d 1184, 1188 (7th Cir.1994). A surgeon exercises a great deal of discretion when deciding how (if at all) to perform an operation, but the fact that an ERISA welfare benefit plan pays for the medical procedure does not make the surgeon a "fiduciary" of the patient and convert all medical-malpractice claims to federal common law under ERISA in the process. What is true at the level of a medical professional is true at the level of a medical practice group such as Carle. Unless the group exercises, not discretion in the abstract, but discretion "in the administration of [the] plan", it is not a fiduciary under ERISA. Lori Pegram, a physician employed by Carle, scheduled Herdrich for an ultrasound examination in Urbana on one day rather than in Bloomington on another; that does not sound like an exercise of discretion "in the administration of [the] plan". Similarly Carle's decision to establish one set of cost-saving incentives rather than another is not an exercise of discretion "in the administration of [the] plan"; it is an exercise of managerial discretion in the administration of Carle's business.

Perhaps it would be possible to read "in the administration of [the] plan" broadly in order to catch all discretionary elements of the HMO structure, but why should courts do this? In order to wipe out HMOs and foreclose the possibility that plan sponsors will choose that structure (or that participants will select it from among options the plan offers)? The panel's opinion sounds very much like this is the objective: its lengthy condemnation of managed care, 154 F.3d at 373-79, otherwise is hard to understand. But ERISA does not tie the plan sponsor's hands on issues of plan design. An employer is free to offer an HMO as an option without objection on fiduciary-duty grounds. Hughes and Lockheed, which put plan-design issues outside the scope of § 1002(21), establish this point. If it is lawful under ERISA for an employer to offer an HMO as a welfare benefit, then it must be lawful for the HMO itself to administer a managed-care system. Boyle v. United Technologies Corp., 487 U.S. 500, 506-07, 108 S. Ct. 2510, 101 L. Ed. 2d 442 (1988), observed that collecting damages from manufacturers of military hardware would, as a practical matter, limit the military's design choices. Just so with ERISA: a plan sponsor's right to adopt an HMO plan as a benefit would not be worth anything if implementing the HMO itself violates ERISA. What the panel has held comes to the same thing--though by a different route--as saying that welfare-benefit plans have a fiduciary duty not to adopt HMO or other managed-care options. If alternatives such as fee-for-service medicine are more expensive, then plan sponsors will be inclined to offer less medical coverage, and participants may be worse off. Clearly the panel thinks that they will be better off, and perhaps they will be. But ERISA allows plan sponsors and participants to choose for themselves. An employer is entitled to offer the combination of fringe benefits that it is willing to pay for; it need not offer the best available medical (or other) services. By stretching the definition of a "fiduciary" under ERISA, the panel has effectively foreclosed a popular option for the delivery of medical care and taken the decision out of private hands, to which ERISA committed it.

If Carle described to State Farm the cost-reduction incentives used by its plan, and State Farm knowingly chose Carle over other providers, then we have a simple plan-design issue. It would defeat the employer's right to specify the benefits conferred by a plan if the dissatisfied employee could turn around and sue the person who delivered those benefits. The only proper question in a suit against the supplier is whether that person did what he promised. Nothing in the panel's opinion suggests that Carle Clinic pulled a fast one on State Farm. Fiduciary duties are vital when contracts are incomplete, but when a contract fully specifies proper behavior, then even a full-fledged trustee need not (indeed, must not) depart from the contractual provisions that the settlor established. See John H. Langbein, The Contractarian Basis of the Law of Trusts, 105 Yale L.J. 625, 657-69 (1995). Carle followed its contract with State Farm and with its subscribers; that is all ERISA requires.

Note the parallel to the fiduciary standard of diversification, to which the panel refers. 154 F.3d at 371-72. Managers of a plan's assets usually have a fiduciary duty to diversify their investments. Yet if the plan sponsor chooses a nondiversified strategy (for example, an employee stock ownership plan), then the trustee administering the ESOP need not diversify; one can't use a fiduciary duty to compel the trustee to disregard a decision consciously made by the sponsor. Even in the private law of trusts a fiduciary may implement the settlor's plan, despite features of that plan that a fiduciary could not adopt on its own (that is, without the settlor's consent). So if Carle is analogous to the trustee, then the settlor's (State Farm's) consent authorizes it to carry out the plan that the settlor approved.

Perhaps this issue boils down to a matter of characterization. If one conceives of particular medical services as the "benefits" under the plan, then Carle serves as the gatekeeper to those benefits, and handling claims for medical benefits defined by a plan is a fiduciary role under ERISA. Massachusetts Mutual Life Insurance Co. v. Russell, 473 U.S. 134, 105 S. Ct. 3085, 87 L. Ed. 2d 96 (1985). But if instead one conceives of the CarleCare HMO system as the benefit promised by the ERISA plan, then Carle is not a "fiduciary." It is just the supplier of medical care, like the surgeon discussed above. Which characterization is best? Herdrich does not allege that State Farm hired Carle to administer a medical plan that offers defined medical procedures as benefits; she alleges, rather, that the benefit State Farm offered is the CarleCare HMO system. And, for reasons I have already discussed, to the extent there is uncertainty about the right way to characterize Carle's role, the court should prefer the characterization that preserves plan sponsors' (and participants') freedom of choice. That means treating the Carle HMO as the benefit, rather than treating Carle as the administrator of the ERISA plan. If the HMO system is the benefit, then Carle is not acting as a fiduciary.

The choice between these characterizations is important--more than enough to justify convening the full court. Most medical care these days is furnished under ERISA plans. Most contemporary welfare benefit plans provide for managed care, through HMOs or other devices, at least as an option. The panel's opinion thus implies that the principal organizational forms through which medical care is delivered today are unlawful. If this conclusion is correct, then the cost-saving achieved by managed care must be abandoned, and the cost of medical care will rise, perhaps substantially.

I recognize that my colleagues in the majority of the panel have expressed their holding as a conclusion about this specific complaint and have written that cost-reduction incentives are not necessarily automatic violations of fiduciary duty. 154 F.3d at 373. But a holding such as this is impossible to cabin, for the plan attacked in this case is an ordinary HMO.

Drawing parallels to the case under consideration, Herdrich sets forth, in the amended third count of her complaint, the intricacies of the defendants' incentive structure. The Plan dictated that the very same HMO administrators vested with the authority to determine whether health care claims would be paid, and the type, nature, and duration of care to be given, were those physicians who became eligible to receive year-end bonuses as a result of cost-savings. Because the physician/administrators' year-end bonuses were based on the difference between total plan costs (i.e., the costs of providing medical services) and revenues (i.e., payments by plan beneficiaries), an incentive existed for them to limit treatment and, in turn, HMO costs so as to ensure larger bonuses. With a jaundiced eye focused firmly on year-end bonuses, it is not unrealistic to assume that the doctors rendering care under the Plan were swayed to be most frugal when exercising their discretionary authority to the detriment of their membership.

154 F.3d at 372 (emphasis in original). If Carle's setup violates ERISA, then all managed care does so, because the allegations in the complaint narrate mundane features of health maintenance organizations. Limiting care to specific locations, limiting referrals to specialists, and using capitation fees (with the possibility of profit from cost-reducing strategies), and reaping for the HMO's owners the benefits of reduced health-care expenditures, are the principal features of HMOs and "preferred provider organizations." Unlike some other HMOs, Carle is owned by its physicians, but I do not think that this makes a legal (or practical) difference. Physicians own much of the stock of HMOs organized as corporations or receive some of its profits as bonuses or salary increments; and no matter the HMO's internal organization, the benefit to a particular physician from a particular treatment decision is minuscule. The effect of holding down costs can be large in the aggregate, but this is so whether the HMO is organized as a corporation or as a partnership. Indeed, it is so whether the organization is an HMO or a law firm. Lawyers owe fiduciary duties to their clients. Can it be that the incentive given by the partnership's reward structure to substitute the services of associates for those of the partners creates a conflict of interest that invariably violates those duties? If the answer is "no" for law firms (and that must be the right answer), it is "no" for HMOs, in stock or partnership form.

Even if all of this is wrong, however, the panel's opinion puts all managed-care systems at risk and commits the court to a long (and I should think unhappy) course of distinguishing "good" managed-care systems from "bad" ones. Assessments of this kind belong to plan sponsors and participants, not to judges. Federal law both recognizes and regulates HMOs. See 42 U.S.C. § 300e. It seems to me unwise and improper for a court to use ERISA to impress a different view of desirable medical care on employers and HMOs alike.