We are once again called upon to determine whether a lawsuit claiming medical negligence is completely preempted by the civil enforcement provision of the Employee Retirement Income Security Act ("ERISA"), 29 U.S.C. § 1132(a). Joseph V. DiFelice, Jr., appeals the order of the United States District Court for the Eastern District of Pennsylvania dismissing his complaint against Aetna/U.S. Healthcare, Inc. ("Aetna") for negligent conduct in regard to his medical treatment for sleep apnea and upper airway obstruction. DiFelice filed suit in state court, alleging that Aetna's instruction to his treating physician that a specially designed tracheostomy tube was "medically unnecessary" and Aetna's insistence that he be discharged from the hospital before his attending physician deemed it appropriate amounted to negligent conduct under state law. Aetna removed the case to federal court on the basis of ERISA preemption and then moved to dismiss the claim. The District Court, relying on our decision in Pryzbowski v. U.S. Healthcare, Inc., 245 F.3d 266 (3d Cir.2001), held that the claim was completely preempted and dismissed it in its entirety. For the reasons that follow, we will affirm in part and reverse in part.

DiFelice participates in an ERISA-governed employee welfare benefit plan that is administered by Aetna, a health maintenance organization ("HMO"). Under the terms of this plan, DiFelice is entitled to certain "Covered Benefits." Unless there is a specific provision for a particular type of treatment, a benefit is only covered if, in the determination of Aetna, it is "Medically Necessary." "Medically Necessary" is a defined term, meaning the service or supply must be "care or treatment as likely to produce a significant positive outcome as, and no more likely to produce a negative outcome than, any alternative service or supply;" must be "related to diagnosis of an existing illness or injury;" may "include only those services and supplies that cannot be safely and satisfactorily provided at home;" and, "as to diagnosis, care and treatment[, must] be no more costly (taking into account all health expenses incurred in connection with the service or supply) than any equally effective service or supply."

In March 2001, DiFelice was diagnosed with "sleep apnea/upper airway obstruction," for which he required a tracheostomy tube.1 His doctor, Dr. Michael Picariello, surgically inserted a tracheostomy tube to eliminate the obstruction, but that tube continually came out. Dr. Picariello then placed an order for a specially designed tube. However, Aetna instructed Dr. Picariello that the special tube was "medically unnecessary." Instead of ordering the special tube, the doctor then inserted a different tube, which caused DiFelice severe pain and resulted in an infection. DiFelice was later admitted to Chester County Hospital for treatment, but, the complaint avers, was thereafter discharged "at Aetna's insistence."2

DiFelice filed a five-count complaint in the Philadelphia Court of Common Pleas against Aetna, his treating physicians, and the hospital. In Count I, he alleged that Aetna negligently interfered with his medical care "by instructing Dr. Picariello that the specially designed tracheostomy tube he deemed necessary was medically unnecessary for [DiFelice] and improperly interfering with Dr. Picariello's medical decision concerning the tracheostomy tube and insisting on [DiFelice's] discharge from the [hospital] ... before his attending physician was planning on discharging [him]." The other counts involved claims against parties other than Aetna. Aetna removed the case to the District Court on the grounds that the claim against it was completely preempted under ERISA and then moved to dismiss. DiFelice opposed the motion to dismiss and moved to remand to state court.

The District Court denied DiFelice's motion to remand and granted Aetna's motion to dismiss as to Count I, and granted the motion to remand on the remaining counts against the other parties. The Court held that the disposition of Count I was "squarely controlled by the Third Circuit's decision in Pryzbowski," in which we held that a claim challenging the "administration of or eligibility for benefits" was completely preempted by section 502(a)(1)(B) of ERISA. Pryzbowski, 245 F.3d at 273. The Court reasoned that the claim against Aetna was completely preempted because DiFelice was challenging Aetna's decision that he was not entitled to the special tube under the Plan, which was entirely a matter of administration, and because Aetna was not actually involved in providing any medical services to DiFelice. DiFelice appeals the District Court's order dismissing Count I.

We have jurisdiction over the District Court's final order pursuant to 28 U.S.C. § 1291, and review the Court's exercise of jurisdiction and order of dismissal de novo. Pryzbowski, 245 F.3d at 268. Aetna bears the burden of proving the federal jurisdiction it seeks. Spectacor Mgt. Group v. Brown, 131 F.3d 120, 127 (3d Cir.1997). In reviewing the complaint, we must accept as true all of DiFelice's factual allegations and draw all reasonable inferences therefrom. Langford v. City of Atlantic City, 235 F.3d 845, 847 (3d Cir.2000).

DiFelice challenges the District Court's removal jurisdiction over Count I of his complaint and asks us to remand to state court. He argues that his negligence action against Aetna is entirely a matter of state law and provides no basis for removal. Aetna counters that DiFelice's negligence action is in fact nothing more than an action to recover benefits due under his plan, and as such is completely preempted by the civil enforcement provision of ERISA, section 502(a).

Under the "well-pleaded complaint" rule, federal question jurisdiction only exists where an issue of federal law appears on the face of the complaint. Franchise Tax Bd. of Cal. v. Const. Laborers Vacation Trust for S. Cal., 463 U.S. 1, 24, 103 S. Ct. 2841, 77 L. Ed. 2d 420 (1983). However, there is an exception to this rule: when a purportedly state-law claim "comes within the scope of [an exclusively] federal cause of action," it "necessarily `arises under' federal law," and is completely preempted. Id.; see also Beneficial Nat'l Bank v. Anderson, ___ U.S. ___, 123 S. Ct. 2058, 2062, 156 L. Ed. 2d 1 (2003) (explaining the preemptive effect of ERISA). The question before us is therefore whether DiFelice's claims of state law negligence on the part of Aetna fall within the scope of the federal causes of action provided in section 502(a) of ERISA, that is, whether the claims could have been brought under that section. If so, then the existence of the federal claim would provide the basis for federal question jurisdiction but at the same time would require dismissal based on complete preemption.

We have had numerous occasions to consider the question of whether a plaintiff's claim against an HMO is covered by section 502(a) and is therefore completely preempted. See, e.g., Pryzbowski, 245 F.3d at 273-75; Lazorko v. Pa. Hosp., 237 F.3d 242, 250 (3d Cir.2000); In re U.S. Healthcare, Inc., 193 F.3d 151, 162-63 (3d Cir.1999); Dukes v. U.S. Healthcare, Inc., 57 F.3d 350, 358 (3d Cir.1995). Determining whether a claim could have been brought under ERISA has proven to be anything but an exact science. In fact, as my colleagues' concurring opinions point out all too well, the exercise seems to have taken on a life of its own, and not a very satisfying or productive life at that. In any event, the statute and our case law chart the path we must follow.

Section 502(a) allows for civil actions to be brought "by a participant or beneficiary ... to recover benefits due to him under the terms of his plan." 29 U.S.C. § 1132(a)(1)(B). The line between an action to recover benefits, which challenges an administrative decision regarding whether a certain benefit is covered under an ERISA plan, and an action alleging negligence or malpractice, which challenges the medical treatment actually provided to a patient, is a blurry one. We have been continually refining the precise test we use in evaluating such claims.

Most recently, in Pryzbowski, we synthesized the discussions contained in our previous opinions and adopted preferable new terminology. We explained that in the past we had attempted to distinguish between claims directed at the quality of benefits received — that is, as to the treatment - which would not fall within section 502(a), and claims that the plans erroneously withheld a quantum of benefits due — focusing on the administration of the plan — which would be completely preempted. Pryzbowski, 245 F.3d at 272. Following this "quality-quantity" rubric, we had held that an allegation that an HMO had failed to exercise reasonable care in providing medical treatment was not preempted, Dukes, 57 F.3d at 358; an allegation that an HMO's policy of discharging newborns within 24 hours after their delivery was essentially a medical determination and was not preempted, In re U.S. Healthcare, 193 F.3d at 163; and an allegation that an HMO's financial disincentives discouraged medical providers from hospitalizing a mentally ill woman was a "quality of care" claim because it occurred in the course of a treatment decision, and was therefore not preempted. Lazorko, 237 F.3d at 250.

However, in Pryzbowski, we found the "quality-quantity" distinction unclear, and suggested that more helpful terminology was utilized by the Supreme Court in Pegram v. Herdrich, 530 U.S. 211, 120 S. Ct. 2143, 147 L. Ed. 2d 164 (2000). Although we recognized that Pegram "concerned fiduciary acts under ERISA and not preemption," we found useful "the distinction made there between eligibility decisions, which turn on the plan's coverage of a particular condition or medical procedure for its treatment," and "treatment decisions, which are choices in diagnosing and treating a patient's condition," and determined that the distinction was "equally applicable for complete preemption analysis." Pryzbowski, 245 F.3d at 273 (quoting Pegram, 530 U.S. at 228, 120 S. Ct. 2143) (emphasis added and internal quotations omitted). We then explained,

Regardless of the language used, the ultimate distinction to make for purposes of complete preemption is whether the claim challenges the administration of or eligibility for benefits, which falls within the scope of § 502(a) and is completely preempted, or the quality of the medical treatment performed, which may be the subject of a state action.

Id.

Using this nomenclature, it was evident to us in Pryzbowski that "a claim alleging that a physician knowingly delayed in performing urgent surgery ... would relate to the quality of care," while on the other hand, "a claim alleging that an HMO declined to approve certain requested medical services or treatment on the ground that they were not covered under the plan would manifestly be one regarding the proper administration of benefits." Id. (citing Jass v. Prudential Health Care Plan, Inc., 88 F.3d 1482, 1488-89 (7th Cir.1996)). However, we were presented there with claims that fell somewhere in between: that an HMO had negligently delayed approval of an out-of-network specialist, and that it had failed to supervise properly its employees to make "thoughtful and reasonable decisions as to healthcare." Id. at 274. Were those claims based on a treatment decision or on a determination as to eligibility for a benefit? We explained that "[i]n analyzing whether a claim falling between the[] two poles is completely preempted, it is necessary to refer to § 502(a)." Id. at 273. Paring the issue down to its essence, we stated that the relevant question must be whether the claim "could have been the subject of a civil enforcement action under § 502(a)." Id. If it could have, then it was a plan benefit claim, and Congress has clearly expressed its intent that the claim be preempted by ERISA. Id. (citing Metropolitan Life Ins. Co. v. Taylor, 481 U.S. 58, 66, 107 S. Ct. 1542, 95 L. Ed. 2d 55 (1987)).

Because the claims in Pryzbowski fell between the two poles, we took extra care to examine the complaint for "artful pleading," to ensure that Pryzbowski was not disguising an eligibility claim that could have been brought under ERISA as a state law negligence claim. We explained that, although the claim might be "ostensibly directed at the provision of medical treatment," we needed to "look beyond the face of the complaint to determine whether [Pryzbowski had] artfully pleaded his suit so as to couch a federal claim in terms of state law." Id. at 274 (quoting Jass, 88 F.3d at 1488); see also Franchise Tax Bd., 463 U.S. at 22, 103 S. Ct. 2841 ("[A] plaintiff may not defeat removal [to federal court] by omitting to plead necessary federal questions in a complaint."). The ultimate question was whether, when the basis for the claim was properly understood, the claim fell under ERISA.

After carefully examining Pryzbowski's complaint for the true bases of his claims, we held that his claims were completely preempted. First, regarding the delayed approval, we concluded that underlying the HMO's allegedly negligent activities was a policy decision regarding payment to and approval of out-of-network specialists, a decision that fell "within the realm of the administration of benefits." Pryzbowski, 245 F.3d at 273. We explained that this claim could have been brought under ERISA because "[h]ad Pryzbowski sought to accelerate [the HMO]'s approval of the use of out-of-network providers, she could have sought an injunction under § 502(a) to enforce the benefits to which she was entitled under the plan." Id. at 273-74. Further, her claim that the HMO had "failed to properly hire, train, and supervise its employees to make thoughtful and reasonable decisions as to healthcare" was also preempted because, reading behind the artful "medical negligence" pleading, the complaint did not allege that the HMO or its employees had actually engaged in any medical treatment. Id. at 274. Because the HMO's only role was in administering Pryzbowski's benefits, it could not possibly havebeen negligent in providing treatment. Id. Unlike the situation in which an HMO fills dual roles as an administrator of benefits and a provider of services, and might therefore actually engage in medical treatment, the HMO there was acting solely as an administrator. Id.

Pryzbowski thus instructs us to determine whether a claim is preempted under section 502(a) by first examining whether the claim falls at either of the two poles, entirely treatment or entirely administrative. If based solely on a medical treatment decision, then the claim is not preempted. If based on an HMO administrator's eligibility decision, then the claim is preempted. In the more difficult situation in which the claim falls somewhere in between, we must scrutinize the complaint for "artful pleading," and then refer to section 502(a) itself and determine whether the actual alleged wrongdoing underlying the cause of action could have formed the basis of a suit under that section.

As discussed more fully below, when we apply the Pryzbowski framework to the complaint before us, we conclude that DiFelice's claim that Aetna "interfered with" his medical treatment by declaring the special tube "medically unnecessary" is preempted by ERISA because it could have been brought as an action under section 502(a). However, because it appears that DiFelice's claim that he was discharged "at the insistence of Aetna" does not rest on any discharge policy set forth in the Plan, or any agreed benefit, it would not be encompassed within the relief available under section 502(a) and is therefore not completely preempted.

We will first examine DiFelice's claim that Aetna interfered with Dr. Picariello's medical decision regarding the special tube. Under Pryzbowski, the first question is whether Aetna's "medical necessity" determination is clearly either a medical treatment decision or an eligibility decision. DiFelice has couched this claim in terms of Aetna's negligent interference with his care, which seems to imply that Aetna itself engaged in medical treatment. However, DiFelice's complaint does not include any allegation that Dr. Picariello was an agent of Aetna, that Aetna did not exercise reasonable care in monitoring Dr. Picariello's care, or that Aetna itself provided medical treatment; rather, his claim rests on Aetna's "instruction" to Dr. Picariello "that the specially designed tracheostomy tube he deemed necessary was medically unnecessary," a direct reference to the "medical necessity" determination called for in the Plan. Looking behind DiFelice's use of language sounding in negligence, he is alleging that Aetna wrongfully denied him coverage for the special tube. Thus, the complaint has aspects of treatment and coverage. That is, in making its determination, Aetna necessarily had to exercise some medical judgment, i.e., it had to determine whether the special tube was "as likely to produce a significant positive outcome as, and no more likely to produce a negative outcome than, any alternative service or supply, ... [was] related to diagnosis of an existing illness or injury, ... [and was] no more costly (taking into account all health expenses incurred in connection with the service or supply) than any equally effective service or supply." However, here there is no allegation that Aetna actually provided the medical care, and Aetna's use of medical judgment could only have led to an eligibility, not a treatment, decision.

Because the decision here was in some sense both a medical treatment and an eligibility decision, thus falling between the two poles discussed in Pryzbowski, we must refer to section 502(a) and determine whether DiFelice's claim regarding the tube could have been the subject of a civil enforcement action under ERISA. Pryzbowski, 245 F.3d at 273. Clearly, it could have been. DiFelice could have challenged Aetna's "medical necessity" determination by filing a claim under 502(a)(1)(B) "to recover benefits due to him under the terms of his plan," and arguing that the special tube was in fact "medically necessary," and was therefore a "covered benefit." He could have requested an injunction forcing Aetna to pay for the special tube, or alternatively, paid for the tube himself and then later filed an action for reimbursement. Numerous ERISA participants have in fact brought such actions challenging their HMO's "medical necessity" determinations and seeking to recover benefits they alleged were due under their plans. See, e.g., Mario v. P & C Food Mkts., Inc., 313 F.3d 758, 762-63 (2d Cir.2002) (reviewing claim under section 502(a) challenging HMO's determination that sex change operation was not "medically necessary"); Fritcher v. Health Care Serv. Corp., 301 F.3d 811, 814-15 (7th Cir.2002) (reviewing claim under section 502(a) challenging HMO's determination that custodial care was not "medically necessary"); Kopicki v. Fitzgerald Auto. Family Employee Benefits Plan, 121 F. Supp. 2d 467, 480 (D.Md.2000) (granting preliminary injunction to prevent HMO from denying preauthorization for cancer treatment it had deemed not "medically necessary"); see also Heasley v. Belden & Blake Corp., 2 F.3d 1249, 1253 (3d Cir.1993) (reviewing claim under section 502(a) challenging an HMO's determination that a liver/pancreas treatment was an "experimental procedure"). DiFelice's claim falls squarely within this jurisprudence. Because DiFelice's claim that Aetna improperly deemed his special tube to be "medically unnecessary" could have been brought under section 502(a), it is completely preempted by ERISA. We will therefore affirm the District Court's exercise of removal jurisdiction and its order dismissing Count I as to Aetna's conduct regarding the tracheostomy tube.

DiFelice urges that this result is inconsistent with the Supreme Court's decision in Pegram and our decisions in U.S. Healthcare and Lazorko. We disagree.

In Pegram, the Supreme Court answered the question whether an HMO may be liable for a breach of fiduciary duty when its physician owners make "mixed eligibility and treatment decisions." Pegram, 530 U.S. at 229, 120 S. Ct. 2143. There, a physician, who was also an owner of the HMO that covered her patient, waited to order an ultrasound for the patient's inflamed appendix, and the appendix ruptured. Id. at 215, 120 S. Ct. 2143. The patient sued the HMO for breach of fiduciary duty, alleging that the HMO created an incentive for the physicians to make decisions in their own financial interests, rather than in the exclusive interests of the plan participants. Id. The Court held that HMOs do not act as "fiduciaries" as envisioned by ERISA when their physician owners make decisions that touch both on the patient's medical treatment and the patient's eligibility for benefits under the plan. Id. at 218, 120 S. Ct. 2143.

The Court first set forth a framework for understanding the kinds of acts that physician owners acting on an HMO's behalf might undertake. On the one hand are "pure `eligibility decisions'" turning on the plan's coverage for a particular medical treatment, and on the other are "treatment decisions," choices about how to go about diagnosing and treating a patient's condition. Id. at 228, 120 S. Ct. 2143. In between are situations in which the "eligibility decision and the treatment decision [are] inextricably mixed," that is, when an eligibility decision necessarily rests on the "physicians' judgments about reasonable medical treatment." Id. The Court found that it was presented with just such a "mixed" decision: the physician owner had determined that the patient's "condition did not warrant immediate action; the consequence of that medical determination was that [the HMO] would not cover immediate care." Id.

Having concluded that the decision before it was a "mixed" decision, the Court went on to hold that an HMO's physician owners do not act as fiduciaries under ERISA when making such decisions. Id. at 231, 120 S. Ct. 2143. The Court focused on the nature of "fiduciaries," explaining that these mixed decisions are not "fiduciary in nature," and bear "only a limited resemblance to the usual business of trustees." Id. The Court feared that a contrary result would open the floodgates to malpractice suits against HMOs and individual physicians under the guise of ERISA breach of fiduciary duty claims, and would erode the distinction between state malpractice and federal ERISA actions. Id. at 235-36, 120 S. Ct. 2143.

Although, as we noted above, Pegram set forth helpful terminology for preemption analysis, see Pryzbowski, 245 F.3d at 273, the Court's holding that a "mixed" determination made by a physician owner does not subject an HMO to liability for breach of fiduciary duty does not translate to, or govern in, the preemption context. Rather, Pegram sets a standard for when liability is to be imposed on individuals acting in a fiduciary capacity. It does not presume to encompass ERISA claims enforcement as such. In fact, the Pegram Court specifically stated that it was not discussing the standards governing a claim that a patient had been wrongfully denied benefits due under a plan nor the interaction between such a claim and state law causes of action. Pegram, 530 U.S. at 229 n. 9, 120 S. Ct. 2143; accord Roark v. Humana, Inc., 307 F.3d 298, 308 (5th Cir.2002) (stating that the Supreme Court has not decided whether section 502(a)(1)(B) preempts a medical malpractice claim involving "mixed decisions," but holding under Fifth Circuit law that it does not). In fact, it could be argued that allowing plaintiffs to do an end-run around ERISA by permitting them to couch plan decisions that have some impact on treatment in negligence terms would have a similar effect of undermining ERISA as was feared by the Court in Pegram. We remain convinced that, after Pryzbowski, we look at what decisions are subject to enforcement, which is a radically different inquiry from when can an HMO be sued for breach of fiduciary duty.3

We are also not persuaded that U.S. Healthcare and Lazorko — both of which pre-date Pryzbowski and rely on the "quality-quantity" distinction — compel a different result. In U.S. Healthcare, we held that a suit against an HMO challenging its policy of pre-certifying a twenty-four hour discharge of mother and newborn was not preempted by ERISA because it went to the quality of the health care provided. U.S. Healthcare, 193 F.3d at 162. Significantly, we noted that the plaintiffs did not allege "a failure to provide or authorize benefits under the plan," nor did they claim "that they were denied any of the benefits that were due under the plan." Id. Rather, they alleged that the HMO was negligent in adopting the discharge policy and in supervising the physicians with whom they contracted for services, and that the HMO's incentive structure adversely affected the physicians' medical judgment regarding when newborns should be discharged. The plaintiffs were not seeking to enforce benefits they thought were due to them under the plan; they were challenging the discharge policy itself and the HMO's actions "in its role as a provider or arranger of medical services," not in its role as administrator of benefits. Id. at 163. The HMO's "policy and incentive structure were such that the [patients] never had the option of making an informed decision as to whether to pay for the hospitalization themselves, as would occur in a situation in which coverage is sought and denied." Id. Although we analyzed the claims in U.S. Healthcare under the "quality-quantity" rubric, our holding squares with the preemption framework we later set forth in Pryzbowski. Because the plaintiffs in U.S. Healthcare were not suing for any benefits due under the plan, but rather were challenging the plan itself, they could not have sued under section 502(a), and therefore, even under Pryzbowski, their claims would not have been preempted.

Lazorko involved a similar issue. There, we held that a claim that financial disincentives imposed by an HMO discouraged medical providers from hospitalizing a patient who later committed suicide was a "quality of care" claim and therefore not preempted. Lazorko, 237 F.3d at 249-250. We noted that Lazorko was not claiming that the HMO plan was supposed to include hospitalization, but rather that her doctor was influenced in his decision-making by the incentive structure. Id. As in U.S. Healthcare, the patient could not have sued under ERISA because she never had the option of seeking continued hospitalization — her doctor, as influenced by the HMO policy, did not offer it to her. Id.

Unlike in U.S. Healthcare and Lazorko, here, DiFelice is suing based on denial of a benefit he claims he was due under the Plan. He is not claiming that Aetna negligently adopted a particular policy regarding tracheostomy tubes or imposed an incentive structure that interfered with his physician's independent medical judgment. Rather, he claims that the special tube was not "medically unnecessary," and that Aetna wrongfully determined that it was, a claim that is based on the language of the Plan and pertains to the nature of benefits provided. Unlike the plaintiffs in U.S. Healthcare and Lazorko, DiFelice sought and was denied coverage for a benefit, and could have paid for the benefit himself and sued under section 502(a) for reimbursement.

Because under our most recent controlling precedent, Pryzbowski, DiFelice's claim that Aetna was negligent in determining that the special tube was "medically unnecessary" could have been the subject of a suit under section 502(a) for benefits due under the Plan, his claim is preempted by ERISA.4 We will therefore affirm the District Court's exercise of removal jurisdiction and subsequent dismissal of the claim with respect to the tube.

Count I also includes a claim that Aetna improperly interfered with DiFelice's medical treatment by "insisting on [his] discharge from the [hospital] ... before his attending physician was planning on discharging [him]." The District Court did not address this aspect of the claim in its order dismissing the complaint.

Unlike his claim regarding the tracheostomy tube, DiFelice does not allege that Aetna deemed the hospital stay to be "medically unnecessary" and therefore not covered by the Plan. The claim also does not appear to rely on a discharge policy in the Plan or any agreed benefit. Indeed, it is difficult to tell from DiFelice's vague pleadings what precisely he is alleging that Aetna did or the form this "insistence" took. If Aetna in some way forced the hospital to discharge him, or imposed financial incentives like those in U.S. Healthcare and Lazorko that unduly affected his physician's judgment, then DiFelice has pled a negligence cause of action that is not preempted by ERISA.

At the dismissal stage, it was Aetna's burden to prove federal jurisdiction by proving that this is an ERISA claim. See Spectacor Mgt. Group, 131 F.3d at 127. There is nothing apparent from the pleadings that would foreclose DiFelice from being able to prove that the discharge decision was not a plan benefit. Unlike his claim that the tube was erroneously determined to be "medically unnecessary" under the Plan, this claim on its face is not plan-related. Therefore, because DiFelice's claim of "insistence" on the discharge could give rise to state law negligence liability, we hold that it is not completely preempted by section 502(a) of ERISA.

Aetna argues that, even if DiFelice's hospital discharge claim is cognizable as a state law cause of action, we should uphold the District Court's dismissal of that claim on the alternative ground that it is nonetheless expressly preempted by virtue of section 514 of ERISA, 29 U.S.C. § 1144(a), which provides that ERISA "shall supersede any and all State laws insofar as they may now or hereafter relate to any employee benefit plan" covered by the statute. However, "[u]nlike the scope of § 502(a)(1)(B), which is jurisdictional and creates a basis for removal to federal court, § 514(a) merely governs the law that will apply to state law claims, regardless of whether the case is brought in state or federal court." Lazorko, 237 F.3d at 248. As we have determined that the hospital discharge claim is not preempted by section 502(a), we do not have jurisdiction over that claim. Rather, the question of whether the hospital discharge claim will be controlled by federal law pursuant to section 514 must be decided by the District Court on remand, should it choose to exercise supplemental jurisdiction under 28 U.S.C. § 1367(a), or by the state court. See Pryzbowski, 245 F.3d at 276 (where District Court has jurisdiction over one claim by virtue of preemption under section 502(a), it has discretion to decide whether to exercise supplemental jurisdiction over claims arising from the same factual predicate or remand to state court).

Because DiFelice's claim that Aetna interfered with his medical treatment by finding the special tracheostomy tube to be "medically unnecessary" could have been brought under section 502(a) of ERISA for the recovery of benefits due under his plan, it is completely preempted. We will therefore affirm the order of the District Court dismissing that claim. However, because Aetna has not shown that DiFelice's claim that Aetna interfered with his medical treatment by insisting on his discharge from the hospital is based on any plan benefit, that claim is not completely preempted. We will therefore reverse the District Court's order dismissing that claim and remand for further proceedings.

BECKER, Circuit Judge, concurring.

I am satisfied that Judge Rendell's opinion reaches the correct result under our governing caselaw. While I thus join in her opinion and in the judgment, I write separately to add my voice to the rising judicial chorus urging that Congress and the Supreme Court revisit what is an unjust and increasingly tangled ERISA regime.

Congress enacted ERISA in 1974 "to promote the interest of employees and their beneficiaries in employee benefit plans." Shaw v. Delta Air Lines, Inc., 463 U.S. 85, 90, 103 S. Ct. 2890, 77 L. Ed. 2d 490 (1983) (surveying Congressional statements of purpose). However, with the rise of managed care and the Supreme Court's series of decisions holding preempted any action for damages against HMOs, ERISA has evolved into a shield that insulates HMOs from liability for even the most egregious acts of dereliction committed against plan beneficiaries, a state of affairs that I view as directly contrary to the intent of Congress. Indeed, existing ERISA jurisprudence creates a monetary incentive for HMOs to mistreat those beneficiaries, who are often in the throes of medical crises and entirely unable to assert what meager rights they possess.

Lower courts have struggled to maintain some semblance of equity notwithstanding the enormous breadth of the preemption test, one that turns on whether the claim is "related to" a benefit plan, inter alia, by identifying exceptions to § 514 preemption, such as that for medical malpractice liability. And in terms of the remedial scope of § 502, they have struggled to make sense out of the distinction between eligibility decisions (which are preempted) and medical decisions (which are not), a hopeless endeavor, as I shall explain. Unfortunately, the price of all this has been descent into a Serbonian bog1 wherein judges are forced to don logical blinders and split the linguistic atom to decide even the most routine cases.

ERISA was enacted in 1974 to address the increasingly-apparent insecurity of workers' vested pension funds, a problem that gained national recognition through such notorious events as the Studebaker plant shutdown in 1963, which caused approximately 4,400 workers to lose their pensions. See generally Michael S. Gordon, Overview: Why Was ERISA Enacted?, U.S. Senate, Special Comm. on Aging, The Employee Retirement Income Security Act of 1974: The First Decade 6-25 (Information Paper) (1984). Prior to ERISA, workers were entitled only to the assets in the pension plan or to nominal pension benefits, whichever was lower; of course, it was entirely possible for a plan to have zero assets, as happened with some frequency when firms closed shop. ERISA contained a raft of provisions designed to protect plan participants against negligent or malfeasant plan managers. For example, it created the Pension Benefit Guarantee Corporation ("PBGC"), an insurer akin to the Federal Deposit Insurance Corporation, to protect against employer insolvency.

The "benefit plans" in need of protection, however, were of two distinct types — pension and welfare — and substantially different policy concerns animated Congress's reform in each area. Pension plans, to which employees contribute over the course of their careers and rely upon to provide retirement income, were thought to present far greater opportunity for mismanagement and underfunding. Unlike welfare benefit plans, pension plans accrue substantial funds that must be invested with prudence, and they must be able to survive continuously changing circumstances. Title 1 of ERISA therefore imposed on pension plan managers comprehensive reporting, disclosure, vesting, minimum funding, and fiduciary duty requirements.

Welfare benefit plans, which include medical, surgical, sickness, accident, disability, death, unemployment, and similar programs, posed a very different set of challenges. Unlike pension plans, welfare benefit plans operate on a "pay as you go" basis, and generally do not entail long-term financial commitments. See John H. Langbein and Bruce A. Wolk, Pension and Employee Benefit Law at 176 (3d ed.2000). ERISA's framers therefore saw no need to impose vesting requirements on such plans. Similarly, minimum funding requirements are intended to guard against plan default by the plan sponsor on its long-term commitments to plan beneficiaries, a minor concern where no substantial funds accrue. The framers thus exempted welfare benefit plans from ERISA's funding requirements as well.

The result is that welfare benefit plans are far less regulated than pension plans. They are subject to ERISA's reporting, disclosure, fiduciary, and remedial rules — those that govern procedure — but exempt from the substantive vesting and funding requirements, which are meant to guarantee a certain level of expected benefits. Congress's relative lack of concern with substantive regulation of welfare plans is clear from the Supreme Court's statement that "ERISA does not mandate that employers provide any particular benefits, and does not itself proscribe discrimination in the provision of employee benefits." Shaw, 463 U.S. at 91, 103 S. Ct. 2890. Indeed, courts have held that the absence of a vesting provision allows employers to amend their plans virtually at will, even in discriminatory fashion. See, e.g., Confer v. Custom Engine Co., 952 F.2d 41, 43 (3d Cir.1991) (noting a plan sponsor may change benefits prospectively by formal written notice); McGann v. H & H Music Co., 946 F.2d 401 (5th Cir.1991) (upholding an employer's amendment of its plan to exclude coverage for AIDS treatment).

Despite ERISA's relatively light regulation of welfare benefit plans, it is clear that the legislation provided substantially greater safeguards for both pension and welfare plan beneficiaries than had previously existed. But ERISA also contained a crucial concession to plan sponsors in the form of § 514(a), an express preemption provision mandating that Titles I and IV of ERISA, which impose the regulations discussed above, "shall supersede any and all State laws insofar as they may now or hereafter relate to any employee benefit plan." Although this section is subject to an important exception that allows states to impose laws regulating insurance, see § 514(b)(2)(A), its breadth is striking. The Supreme Court explains that Congress intended

to ensure that plans and plan sponsors would be subject to a uniform body of benefits law; the goal was to minimize the administrative and financial burden of complying with conflicting directives among States or between States and the Federal Government ..., [and to prevent] the potential for conflict in substantive law ... requiring the tailoring of plans and employer conduct to the peculiarities of the law of each jurisdiction.

Ingersoll-Rand Co. v. McClendon, 498 U.S. 133, 142, 111 S. Ct. 478, 112 L. Ed. 2d 474 (1990). This view is premised on the statement of Rep. John Dent, a sponsor of the Act in the House of Representatives, who declared that its purpose was to "eliminat[e] the threat of conflicting and inconsistent State and local regulation." 120 Cong. Reg. 29197 (1974).

In a meaningful sense, Congress's decision to federalize pension and employee benefit law appear to have assisted plan beneficiaries as much as plan sponsors. No employer is required to offer a pension or welfare benefit plan to its employees, and were Congress to have established ERISA's regulations as merely a federal baseline and allowed states to supplement it as they saw fit, i.e., had ERISA not preempted state law, some employers might have declined to offer such plans rather than deal with the compliance costs of tailoring them to individual jurisdictions. At a minimum, it is likely that any compliance costs would have been reflected in a lower level of benefits for plan participants, which of course would undermine the purpose of ERISA. From an ex ante perspective, therefore, it might have been reasonable to view even § 514(a) as furthering the interests of plan beneficiaries.

In practice, however, nothing has been further from the truth — ERISA generally, and § 514(a) particularly, have become virtually impenetrable shields that insulate plan sponsors from any meaningful liability for negligent or malfeasant acts committed against plan beneficiaries in all too many cases. This has unfolded in a line of Supreme Court cases that have created a "regulatory vacuum" in which virtually all state law remedies are preempted but very few federal substitutes are provided. In these cases, which began with Alessi v. Raybestos-Manhattan, Inc., 451 U.S. 504, 101 S. Ct. 1895, 68 L. Ed. 2d 402 (1981), the Court initially employed textualist interpretations of § 514(a) to give ERISA a staggeringly broad preemptive scope. Alessi itself was a relatively straightforward case in which a unanimous Court held preempted a New Jersey law that eliminated one method for calculating pension benefits, reasoning that, whatever the law's purpose, it clearly related to pension plans. Id. at 524, 101 S. Ct. 1895. Although Alessi itself was innocuous enough, it established a precedent of determining ERISA's preemptive scope by searching the phrase "relates to" for concrete meaning, an approach that would eventually be chimerical.

In Shaw v. Delta Air Lines, 463 U.S. 85, 103 S. Ct. 2890, 77 L. Ed. 2d 490 (1983), another unanimous decision, two New York statutes required health and disability plans to treat pregnancy the same as other nonoccupational disabilities at a time when federal employment discrimination statutes did not. In a landmark passage, the Court stated:

We have no difficulty in concluding that the [New York laws] "relate to" employee benefit plans. The breadth of § 514(a)'s pre-emptive reach is apparent from that section's language. A law "relates to" an employee benefit plan, in the normal sense of the phrase, if it has a connection with or reference to such a plan.

Id. at 96-97, 103 S. Ct. 2890 (emphasis added). Although the Court left the door ajar by cautioning that "[s]ome state actions may affect employee benefit plans in too tenuous, remote, or peripheral a manner to warrant a finding that the law `relates to' the plan," id. at 100, n. 21, 103 S. Ct. 2890, the "connection with or reference to" language became the test for future cases.

The Court applied the passage in unanimous holdings in Metropolitan Life Ins. Co. v. Massachusetts, 471 U.S. 724, 105 S. Ct. 2380, 85 L. Ed. 2d 728 (1985) (holding that a state statute mandating mental health coverage in group health insurance policies fell within § 514(a)'s preemptive scope, but was saved by ERISA's insurance exception), and Pilot Life Ins. Co. v. Dedeaux, 481 U.S. 41, 107 S. Ct. 1549, 95 L. Ed. 2d 39 (1987) (holding preempted a state cause of action against an insurer for bad-faith denial of a claim). By the mid-1990s, however, it had become clear that the "relates to" standard was one without conceptual limits, and given the Court's general "assumption that the historic police powers of the States were not to be superseded by the Federal Act unless that was the clear and manifest purpose of Congress," Rice v. Santa Fe Elevator Corp., 331 U.S. 218, 230, 67 S. Ct. 1146, 91 L. Ed. 1447 (1947), it balked at the specter of a preemptive vortex that could swallow virtually any state remedial law. It abruptly reversed course in New York State Conference of Blue Cross & Blue Shield Plans v. Travelers Ins. Co., 514 U.S. 645, 115 S. Ct. 1671, 131 L. Ed. 2d 695 (1995), where it unanimously noted that, "[i]f `relate to' were taken to extend to the furthest stretch of its indeterminacy, then for all practical purposes pre-emption would never run its course, for really, universally, relations stop nowhere." Id. at 655, 115 S. Ct. 1671 (internal citation omitted). It recognized that it "simply must go beyond the unhelpful text ... and look instead to the objectives of the ERISA statute as a guide to the scope of the state law that Congress understood would survive." Id. at 656, 115 S. Ct. 1671.

As discussed above, Congress's intent behind § 514(a) preemption was to ensure that plan sponsors would be subject to uniform law. See Ingersoll-Rand, 498 U.S. at 142, 111 S. Ct. 478. One critical aspect of this uniformity concerns § 502, ERISA's civil enforcement provision. In Pilot Life, the Court concluded that Congress intended the remedies in § 502 to be the exclusive remedies for violations of rights guaranteed under ERISA. It explained that "the pre-emptive force of § 502(a) was modeled on the exclusive remedy provided by § 301 of the Labor Management Relations Act.... Congress was well aware that the powerful pre-emptive force of § 401 of the LMRA displaced" all state law claims, "even when the state action purported to authorize a remedy unavailable under the federal provision." 481 U.S. at 52, 53, 107 S. Ct. 1549. In other words, § 514(a) preempts state causes of action to enforce ERISA-guaranteed rights even when § 502 provides no substitute federal cause of action. This "regulatory vacuum" creates situations in which plan beneficiaries have little or no recourse for even the most egregious violations of their rights, for the remedies contained in § 502 are incapable of making victims whole; indeed, in many cases they actually create incentives for HMOs to mistreat their plan participants.

Section 502 contains two subsections that address a participant's right to recover for wrongs committed against the participant personally (as opposed to those committed against the plan itself, which are considered violations of the sponsor's fiduciary duty). The first, § 502(a)(1)(B), authorizes the participant or beneficiary to bring a civil action "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." It is through this provision that DiFelice might have sought an injunction to compel Aetna to cover the specialized tracheostomy tube. The second, § 502(a)(3), authorizes a participant, beneficiary, or fiduciary to seek equitable remedies — injunctive relief against "any act or practice which violates" ERISA or the plan terms, or "other appropriate equitable relief (i) to redress such violations or (ii) to enforce any provisions" of ERISA or the plan.

Although these provisions seem comprehensive at first glance — they allow recovery of "benefits due" and empower a participant to "enforce his rights" and seek "appropriate equitable relief" — they in fact operate to leave participants almost completely at the mercy of HMOs. The first section, 502(a)(1)(B), by its plain language only allows plan participants to seek the benefits to which they are contractually entitled, even when those benefits have been denied in bad faith and despite the fact that the participants most in need of this section are often the ones least able to take advantage of it. A plan participant whose claim is denied by an HMO's utilization review board — Mr. DiFelice, for example, see infra — is often in the throes of a life-or-death medical crisis, hardly a feasible time to retain counsel and prosecute an injunctive lawsuit. The costs of such suits are likely to be immense, and ERISA provides for attorney fees, if at all, only after the action concludes. See § 502(g)(1). Even if a participant ultimately prevails against his insurer, it will frequently be the participant's estate that reaps the meager reward.

In many areas of law contingency fee structures are used to overcome a litigant's initial impecuniosity. But any possibility of using contingency fees in this context is undermined by ERISA preemption, for a string of Supreme Court cases has interpreted ERISA to disallow any recovery of compensatory or punitive damages. See Langbein & Wolk, Pension and Employee Benefit Law at 770-75. In Massachusetts Mutual Life Ins. Co. v. Russell, 473 U.S. 134, 105 S. Ct. 3085, 87 L. Ed. 2d 96 (1985), the plaintiff sued under § 502(a)(2), alleging that the improper processing of her benefit claim exacerbated her condition and entitled her to compensatory and punitive damages under § 409(a), which, inter alia, authorizes "such ... equitable or remedial relief as the court may deem appropriate." The Court, however, concluded that any recovery under § 409 must inure to the plan rather than the beneficiary. The Court then explained in dictum that § 502(a)(1)(B), the analogous section allowing for a participant's recovery, says "nothing about the recovery of extra-contractual damages." Id. at 144, 105 S. Ct. 3085. In an oft-cited passage, it explained.

The six carefully integrated civil enforcement provisions found in § 502(a) of the statute as finally enacted [] provide strong evidence that Congress did not intend to authorize other remedies that it simply forgot to incorporate expressly. The assumption of inadvertent admission is rendered especially suspect upon close consideration of ERISA's interlocking, interrelated, and interdependent remedial scheme, which is in turn part of a comprehensive and reticulated statute.

Id. at 146, 105 S. Ct. 3085 (internal citation omitted).

Although Russell's narrow holding left open the possibility that extracontractual damages are recoverable under § 502(a)(3)'s "appropriate equitable relief" provision, the Court's subsequent decision in Mertens v. Hewitt Associates, 508 U.S. 248, 113 S. Ct. 2063, 124 L. Ed. 2d 161 (1993), foreclosed any such hopes. In Mertens, participants sued the fiduciaries of a failed plan, alleging breach of fiduciary duty and seeking monetary relief, which they characterized as equitable. Petitioners argued that, "[a]lthough a beneficiary's action to recover losses resulting from a breach of duty superficially resembles an action at law for damages, ... such relief traditionally has been obtained in the courts of equity and therefore is, by definition, equitable relief." Id. at 255-56, 113 S. Ct. 2063.

The Supreme Court acknowledged that ERISA's roots lie in the common law of trusts, see Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 110-11, 109 S. Ct. 948, 103 L. Ed. 2d 80 (1989), that courts of equity had jurisdiction over trust law, and that monetary damages were available in those courts against the trustee. It concluded, however, that "[s]ince all relief available for breach of trust could be obtained from a court of equity, limiting the sort of relief available under § 502(a)(3) to `equitable relief' in the sense of `whatever relief a common-law court of equity could provide in such a case' would limit the relief not at all." Mertens, 508 U.S. at 257, 113 S. Ct. 2063. Put another way, interpreting "equitable relief" to mean "any relief" would render superfluous the word "equitable." The Court's solution was to interpret "equitable" as referring to types of relief traditionally available in pre-merger courts of equity, that is, injunctions, mandamus, and restitution, but not monetary damages.

The unavailability of extracontractual damages has effects that are perverse. First, as stated above, contingency fees are rendered entirely impractical — precious few lawyers would be willing to undertake a horrendously complex case of uncertain outcome when the greatest potential reward is merely provision of the care that had been contractually promised. Without contingency fees, participants in the midst of medical crises are completely at the mercy of HMOs unless they are fortunate enough to have the financial means to bring a suit for an injunction, a circumstance which is no doubt exceptional. Although it might seem a simple matter to seek an injunction compelling contractually-guaranteed coverage of a procedure, nothing is further from the truth where — as with Mr. DiFelice — the contractual availability of coverage hinges on a highly fact-intensive determination of medical necessity involving accepted standards of care and tolerable levels of risk for the participant's malady. To the extent that participants are unable to seek an injunction compelling coverage, ERISA's remedial scheme is almost entirely illusory.

The second perverse effect is that, at the same time as ERISA makes it inordinately difficult to bring an injunction to enforce a participant's rights, it creates strong incentives for HMOs to deny claims in bad faith or otherwise "stiff" participants. ERISA preempts the state tort of bad-faith claim denial, see Pilot Life, 481 U.S. at 54-56, 107 S. Ct. 1549, so that if an HMO wrongly denies a participant's claim even in bad faith, the greatest cost it could face is being compelled to cover the procedure, the very cost it would have faced had it acted in good faith. Any rational HMO will recognize that if it acts in good faith, it will pay for far more procedures than if it acts otherwise, and punitive damages, which might otherwise guard against such profiteering, are no obstacle at all. Not only is there an incentive for an HMO to deny any particular claim, but to the extent that this practice becomes widespread, it creates a "race to the bottom" in which, all else being equal, the most profitable HMOs will be those that deny claims most frequently.

In sum, ERISA's remedial scheme gives HMOs every incentive to act in their own and not in their beneficiaries best interest while simultaneously making it incredibly difficult for plan participants to pursue what meager remedies they possess, a confounding result for a statute whose original purpose was to protect employees.

Given that ERISA's remedial scheme often provides no remedies, litigants have gone to great lengths to identify state causes of action that are not preempted by ERISA, and courts have generally been sympathetic to their efforts even when ultimately finding their claims preempted. For example, in Andrews-Clarke v. Travelers Ins. Co., 984 F. Supp. 49 (D.Mass. 1997), a clearly frustrated court inveighed:

Under traditional notions of justice, the harms alleged — if true — should entitle [plaintiff] to some legal remedy on behalf of herself and her children against Travelers and Greenspring. Consider just one of her claims — breach of contract. This cause of action — that contractual promises can be enforced in the courts — pre-dates Magna Carta. It is the very bedrock of our notion of individual autonomy and property rights. It was among the first precepts of the common law to be recognized in the courts of the Commonwealth and has been zealously guarded by the state judiciary from that day to this. Our entire capitalist structure depends on it.

Nevertheless, this Court had no choice but to pluck [plaintiff's] case out of the state court in which she sought redress (and where relief to other litigants is available) and then, at the behest of Travelers and Greenspring, to slam the courthouse doors in her face and leave her without any remedy.

Id. at 52-53.

It is no exaggeration to say that the federal courts have struggled mightily to maintain fidelity to ERISA's expansive "relates to" preemption clause while avoiding the wholesale foreclosure of participants' causes of action against their HMOs. But our search for a middle ground has proved to be a judicial snipe hunt, and we are no closer to success today than we were a decade ago. This Court's own decisions illustrate the quagmire in which courts find themselves. In Dukes v. U.S. Healthcare, 57 F.3d 350 (3d Cir.1995), a participant sued his HMO alleging medical malpractice, a state-law tort, and the HMO argued that the claim was preempted by ERISA. We held that the claim was not preempted, reasoning that ERISA draws a distinction between actions challenging the quality of care provided and those claiming that the HMO provided an inadequate quantity of benefits under its plan. As we explained, actions challenging the quantity of benefits received are actions that could be brought under ERISA § 502(a)(1)(B) to "recover benefits due ... under the terms of the plan," and any state law duplicating that remedy is preempted by ERISA § 514(a). Regulation of quality of care, conversely, was found to be "a field traditionally occupied by state regulation," id. at 357, and we did not find "anything in the legislative history, structure, or purpose of ERISA suggest[ing] that Congress viewed § 502(a)(1)(B) as creating a remedy for a participant injured by medical malpractice." Id.

We further explicated this quantity-quality distinction in In re U.S. Healthcare, 193 F.3d 151 (3d Cir.1999), which concerned, inter alia, a plan participant's claim that her HMO was negligent in adopting a policy of presumptively discharging newborn infants within 24 hours. We "recognize[d] that the distinction between the quantity of benefits due under a welfare plan and the quality of those benefits will not always be clear," id. at 162, and explained that the line-drawing difficulty arises in part "because the same HMO may have assumed both the role as a plan administrator and the separate role as a provider of medical services." Id. The policy of discharging newborns after 24 hours was, however, determined to be a function of the "role as provider of medical services" because it was an "essentially medical determination of the appropriate level of care ..., not a claim that a certain benefit was requested and denied." Id. at 162-63. We therefore held that the plan participant's claim "fit[] squarely within the class of claims that we identified in Dukes as involving the quality of care" and was not preempted. Id. at 163.

Finally, in our most recent and important ERISA preemption decision, Pryzbowski v. U.S. Healthcare, Inc., 245 F.3d 266 (3d Cir.2001), harm was allegedly caused by an HMO's delay in approving a procedure to be performed by off-network physicians. There was no question whether the treatment itself was covered — the issue was merely one of eligibility, namely whether an off-network physician could perform the procedure. We took note of the Supreme Court's recent decision in Pegram v. Herdrich, 530 U.S. 211, 120 S. Ct. 2143, 147 L. Ed. 2d 164 (2000), where the Court distinguished between an HMO's eligibility decisions and treatment decisions. Although Pegram directly concerned fiduciary duty rather than preemption, we held that its eligibility-treatment dichotomy was equally applicable in the preemption context; indeed, we considered it another way of viewing the quantity-quality distinction. We concluded that an eligibility decision is an administrative decision, and that any state-law action challenging it is preempted because a participant could instead bring suit under ERISA § 502(a)(1)(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 right to future benefits under the terms of the plan." Conversely, a treatment decision is a medical decision that requires no interpretation of the plan itself, and it is not preempted by ERISA since § 502(a)(1)(B) offers a plaintiff no avenue of relief.

I believe that Dukes, In re U.S. Healthcare, and Pryzbowski were all rightly decided, and that our opinion today reaches the correct conclusion based on those precedents. I write separately, however, to make clear my concern that the opinion masks extraordinary subtleties and complexities of this area of the law that cry out for clarification by the Congress, or, failing that, by the Supreme Court. In fact, I believe that the fundamental distinction upon which federal caselaw currently relies — between quantity and quality decisions, or between eligibility and treatment decisions — is untenable, and that the blurring is becoming more severe, not less. To the extent we insist on categorizing every HMO decision as either an eligibility or a treatment decision, we contort ourselves into parsing terms that are conceptually indistinguishable, and we fail to come to terms with the realities of modern health coverage.

This problem is evident even among relatively "easy" cases. Pryzbowski is perhaps one of the clearest-cut examples of an eligibility decision, for the HMO's only action was designing its plan to utilize network doctors, a clear matter of plan administration. Conversely, when one thinks of "treatment" or "medical" decisions, one envisions a physician and an examination table, a world seemingly far removed from the HMO's administrative offices. Yet even in Pryzbowski, it takes little creativity to recast the "administrative" matter of plan design as a "treatment" or "medical" decision, for the HMO implicitly determined that requiring plan participants to see network physicians would not expose them to any undue health risks. It is impossible to characterize that decision as anything other than partially medical, for it directly affected participants' health. Indeed, the delay occasioned by the HMO's so-called "administrative decision" quite literally caused Ms. Pryzbowski's condition seriously to worsen.

The Supreme Court itself recognized this false dichotomy problem in Pegram, where it observed that many (and possibly most) HMO decisions contain elements of both eligibility and treatment considerations:

[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.... In practical terms, these eligibility decisions cannot be untangled from physicians' judgments about reasonable medical treatment.

530 U.S. at 228-29, 120 S. Ct. 2143.

Some courts have attempted to explain away the evident problems in the "eligibility versus treatment" dichotomy by holding that only HMOs which employ their own physicians can make medical decisions. See, e.g., Rubin-Schneiderman v. Merit Behavioral Care Corp., 163 F. Supp. 2d 227 (S.D.N.Y.2001). Such courts suggest that HMOs which do not directly employ physicians cannot engage in "mixed" decisions because they do not actually treat patients. The Supreme Court itself has cast doubt upon this explanation, stating that decisions are considered "mixed" "not merely because ... treatment and eligibility are made by the same person, the treating physician." Pegram, 530 U.S. at 228, 120 S. Ct. 2143. But this distinction is logically tenuous even absent that precedent, for an HMO that employs no physicians must nevertheless review the independent physicians' recommendations to determine whether they comport with the health plan's requirements.

The case at bar makes evident the impossibility that any such simple rule can adequately reflect the Byzantine complexities of modern-day health care. AETNA's policy explicitly equated its coverage (i.e., eligibility) on an assessment of medical necessity. But the medical necessity clause, which attempts to convert medical decisions into eligibility decisions, compounds the problem I address herein. Indeed, Aetna denied coverage even though Dr. Picariello insisted the specialized tracheostomy tube was critical to DiFelice's health. Tracheotomy is an extremely serious procedure, and the tracheostomy tube, whose uses are described in the margin, is an integral part of the procedure.2 As two prominent medical school professors explain, tracheotomy "is associated with multiple and potentially life-threatening complications even under elective conditions." Eugene N. Myers and Ricardo L. Carrau, Early Complications of Tracheotomy: Incidence and Management, 12 Clinics in Chest Medicine No. 3 at 589 (September 1991). Indeed, "[e]ven with the use of optimal surgical techniques, complications of tracheotomy may occur during the procedure, in the immediate postoperative period, or long after the surgery." Eugene N. Myers and Sylvan E. Stool, Complications of Tracheotomy, Tracheotomy 147 (Churchill Livingstone 1985). They counsel that "[t]he best means of preventing complications whenever they occur [] are paying meticulous attention to detail in the performance of the surgery and performing the tracheotomy as soon as it becomes obvious that the procedure is necessary." Id.

This assessment is borne out by the history of DiFelice's condition. In March of 2001, DiFelice was diagnosed with sleep and upper airway obstruction. His otolaryngologist, Dr. Michael Picariello, attempted several unsuccessful treatments before determining that DiFelice needed a tracheostomy tube. Dr. Picariello surgically placed a tracheostomy tube in July of 2001, but it continually extubated from DiFelice's neck. It was only then that Dr. Picariello determined that DiFelice needed a specially-designed tube to remedy a potentially life-threatening condition.

Aetna, however, overruled Dr. Picariello's expert medical judgment and determined that the tube was in fact not medically necessary, leaving DiFelice with the option of bringing an injunctive suit under ERISA § 502, paying out-of-pocket for the specially-designed tube, or receiving a second standard-shaped tube that Aetna agreed to cover. He opted for the covered tube, but after it was placed, he developed a serious and progressive soft tissue and bone infection that caused him to be admitted to Chester County hospital in October of 2001, following which he was referred to the Hospital at the University of Pennsylvania. There, doctors removed "significant portions" of his bone and tissue to treat the infection, and his pectoral muscle was surgically reconfigured.

Aetna claims that its decision was not medical merely because it was made with an eye to plan language. To me this makes no sense, for Aetna made precisely the same individualized determination of medical necessity as Dr. Picariello. The fact that Aetna is an HMO and Dr. Picariello is an independent physician is entirely irrelevant to the fundamental character of their assessments. My conclusion that Aetna's decision may have been largely medical suggests that perhaps DiFelice's claim should not be preempted. Such a result would follow comfortably under Dukes and In re U.S. Healthcare. I agree with Judge Rendell, however, that Supreme Court precedent compels the conclusion that the proper test is whether a suit is theoretically possible under § 502(a), an approach which leads to preemption under the facts of this case.

Judge Rendell's opinion recognizes that Aetna's decision had a medical component. It declines to characterize it as either eligibility or treatment, and concludes that in such mixed situations, preemption turns on the availability relief under ERISA § 502(a). This resolution at least has the salutary effect of creating a bright-line rule, perhaps the best we can hope for absent intervention from a higher authority that would enable a truly principled jurisprudence. But while this rule is relatively easy to apply, that ease comes at the direct expense of plan participants' welfare. The rule's premise is that HMOs that do not employ their own physicians are solely in the insurance business, that is, they do not provide care and cannot be medically negligent. In the opinion's words, "because there is no allegation that Aetna actually provided the medical care, Aetna's use of medical judgment could only have led to an eligibility, not a treatment, decision." I believe that this statement (with which I do not agree) encapsulates precisely why ERISA's failure to change with the times has rendered it incapable of protecting employees, and why Congress must act to prevent further injustice.

As discussed above, existing Supreme Court precedent holds that ERISA disallows extracontractual damages even in instances of bad faith, an interpretation that gives safe harbor to HMOs that deny claims while also destroying any possibility of participants bringing § 502 actions under contingency fee arrangements. The result is that, in many cases, participants must take HMOs' decisions as law — for example, when Aetna's utilization review board denied coverage for DiFelice's specialized tracheostomy tube, he faced the decision whether to pay for the specialized tube out-of-pocket, whereas appealing the HMO's decision was simply impractical in the face of a medical emergency. In such cases, the critical insight is that the HMO de facto determines a patient's actual treatment along with his eligibility for benefits, for it will be a relatively rare person who is able to pay for invasive procedures out-of-pocket.

Because ERISA creates a system in which an HMO's benefit determination frequently determines the actual treatment a participant receives, it follows directly that HMOs determine quality of care — and make treatment decisions — regardless of whether they actually employ physicians. Put differently, the root of courts' ERISA preemption nightmare is that ERISA forces them to distinguish between eligibility and treatment decisions while providing a remedial structure that makes the two virtually synonymous. For participants, the torment is still greater: ERISA de facto places the HMO in control of the treatment a participant receives, yet it preempts any state-law medical malpractice claim against that HMO and provides that the participant can recover no compensatory, punitive, or wrongful-death damages regardless of its malfeasance.

This situation has arisen because ERISA has failed to evolve to accommodate the rise of HMOs, which did not even exist when ERISA was enacted in 1974. Back then, fee-for-service insurers dominated the health care industry. See Pegram, 530 U.S. at 218, 120 S. Ct. 2143; see also Kent G. Rutter, Democratizing HMO Regulation to Enforce the "Rule of Rescue", 30 U Mich. J.L. Ref. 147, 171 (1996). Insurers had little role in a person's treatment decisions; instead, a participant would visit a doctor or hospital, receive treatment, and the doctor or hospital would bill the insurer. If an insurer refused to pay, the participant could bring suit under ERISA § 502(a) to recover benefits due under the terms of the plan. This was the role Congress envisioned for § 502, and it worked well, for any disagreement with an insurer would occur only after the participant's medical crisis had abated, and in some ways the system created an incentive to provide too much care rather than too little.

Times have changed. Today, approximately 75% of insured American workers receive their health care through some type of "managed care" plan, a designation which includes HMOs. See Andy Miller, Managed Care Savings Noted, Atlanta J. & Const., June 5, 1997, at E3. One hallmark of managed care systems is the utilization review board, which approves or denies coverage for a procedure before the procedure actually takes place. Although a participant may appeal a utilization review board's decision, the prior-approval system is thought to reduce costs to HMOs because participants are likely to choose an inferior (but approved) procedure over a superior procedure for which they might ultimately pay out-of-pocket following an unsuccessful appeal. ERISA is often ill-equipped to deal with the phenomenon of the utilization review board, for the lack of remedies available under § 502, as discussed above, actively encourages HMOs to deny claims. Because these denials now take place before the treatment itself, the effect is a systematic deterioration in the quality of treatment participants receive, all oxymoronically occasioned by a statute "designed to promote the interests of employees and their beneficiaries in employee benefit plans." Shaw, 463 U.S. at 90, 103 S. Ct. 2890.

What is to be done? Not much, absent intervention by Congress or the Supreme Court, for lower courts are bound to follow precedents that lead inexorably to the "availability of § 502 relief" preemption test set forth by the majority opinion in this case. However, several promising avenues exist. One is suggested by the Bipartisan Consensus Managed Care Improvement Act of 1999, H.R. 2723, 106th Cong., 1st Sess. (1999), which passed the House but was watered down in the Senate. See Langbein & Wolk, Pension and Employee Benefit Law at 561-62. That Act would have amended ERISA § 514 by adding a new subsection (e) providing that ERISA shall not:

be construed to invalidate, impair, or supercede any cause of action brought by a participant or beneficiary (or the estate of a participant or beneficiary) under State law to recover damages resulting from personal injury or for wrong death against any person —

(i) in connection with the provision of insurance, administrative services, or medical services by such person to or for a group health [plan], or

(ii) that arises out of the arrangement by such person for the provision of such insurance, administrative services, or medical services by other persons.

The Act would have disallowed punitive damages when the cause of action relates to an "externally appealable decision." It set up such external appeal procedures for denials of benefit claims based on decisions that "the item or service is not medically necessary or appropriate or is investigational or experimental" or "in which the decision as to whether a benefit is covered involves a medical judgment."

Put more concretely, had the Act become law, Aetna would now face possible compensatory damages but not punitive damages, and its determination of medical necessity would be externally appealable. Although the Act passed the House, it did not survive in the Senate, which passed a version of health care reform containing no right to sue. See H.R. 2990, 106th Cong., 1st Sess. (1999). As suggested, the legislation approved by the House is one approach. There are doubtless others.

Even if Congress refuses to act, however, the Supreme Court, in its interpretive capacity, is capable of effecting salutary change in many ways. The Court has no crystal ball, and twenty years ago it could not have foreseen the radical changes that have overtaken the health care system, and the difficulties that its preemption decisions would create. The time might be right to reconsider the string of holdings, epitomized by Russell and Mertens, that rule out the possibility of recovering compensatory damages under ERISA § 503(a)(3). See generally John H. Langbein, What ERISA Means by "Equitable": The Supreme Court's Trail of Error in Russell, Mertens, and Great-West, Yale Law School Center for Law, Economics, and Public Policy, Research Paper No. 269, available online at www.ssrn.com. Professor Langbein persuasively argues: (1) that the Supreme Court erred in interpreting ERISA's language providing for "other appropriate equitable relief" to mean only relief that was traditionally available in courts of equity; and (2) that the better view is that Congress intended "to remedy ERISA wrongs of the sort commonly remedied under trust law," and a core principal of trust law, the "make-whole standard," attempts to "restore[] the victim to the position that he or she would have had had there been no breach of trust."

That is precisely what compensatory damages do, and there is no reason to suspect that Congress intended to base ERISA on the law of trusts while omitting the predicate law's core remedy. I note that I am not alone among federal judges in urging the Supreme Court to reconsider what has become a significant barrier to justice and the realization of Congressional intent. See, e.g., Cicio v. Does, 321 F.3d 83, 107 (2d Cir.2003) (Calabresi, J., dissenting) ("[I]t is not too late for the Supreme Court to retrace its Trail of Error and start over from the beginning, or for Congress to wipe the slate clean."). Cf. Corcoran v. United HealthCare, 965 F.2d 1321, 1336 (5th Cir.1992) ("The characterization of equitable relief as encompassing damages necessary to make the plaintiff whole may well be consistent with the trust law principles that were incorporated into ERISA and which guide its interpretation.").

Another possibility for the Supreme Court, especially given its recent trend toward looking beyond the text of § 514(a) to determine ERISA's preemptive scope, is to reexamine Congress's intent (or lack thereof) with respect to preempting welfare benefit plans. Many scholars have noted that Congress did not carefully consider the scope of preemption when it drafted ERISA. See, e.g., Fick, The Last Article about the Language of ERISA Preemption?, 33 Harv. J. on Legis. 35, 53 (1996). The House bill would have preempted state laws that "relate to the reporting and disclosure responsibilities and fiduciary responsibilities of persons acting on behalf of" ERISA-covered plans, and state laws that "relate to" funding and benefits-vesting provisions of pension plans. H.R. 2, 93d Cong., 1st Sess. (1973), 120 Cong. Rec. 4742 (1974). The Senate version, on the other hand, would have preempted state laws that "relate to the subject matters regulated by this Act." S. 4200, 93d Cong., 1st Sess. (1973), 120 Cong. Rec. 5002 (1974).

In the final joint conference, the Committee adopted the present language, but made it available to the full Congress only ten days before the bill was enacted, and said little about the change. See Metropolitan Life Ins. Co., 471 U.S. at 745 n. 23, 105 S. Ct. 2380. There is scant reason to believe that the resulting language was fully considered by the entire deliberative body; indeed, those who paid attention to the issue opined that § 514(a) was provisional. Section 3022 mandated the creation of a Joint Pension Task Force to study the practical effect and desirability of preemption, and Senator Jacob Javits, a sponsor of the legislation, said that "the desirability of further regulation — at either the State or Federal level — undoubtedly warrants further attention." 120 Cong. Rec. 29,942 (1974) (remarks of Sen. Javits). Unfortunately, the Task Force never came into existence, and no further regulation was forthcoming.

The evidence suggests that Congress did not carefully consider whether the scope of preemption should reflect the different degrees of federal regulation of pension plans and welfare benefit plans. See Fisk, The Last Article about the Language of ERISA Preemption?, 33 Harv. J. on Legis. at 56. In my view, section 514(a)'s broad preemptive scope is sensible with regard to pension plans, for federal law fully displaces state law and provides vesting, requirements, minimum funding requirements, and a raft of other employee safeguards. However, to me, it makes much less sense with respect to welfare plans. As discussed supra, Congress exempted welfare benefit plans from most of ERISA's substantive regulations, such as its vesting and minimum funding requirements.

As I see it, it is unlikely that Congress intentionally created this so-called, "regulatory vacuum," in which it displaced state-law regulation of welfare benefit plans while providing no federal substitute. The more likely explanation is that Congress merely intended to create minimum safeguards to protect the financial integrity of welfare benefit plans while stopping short of federalizing the entire remedial regime, especially in light of what was a workable state-law remedial system. Congress's failure to distinguish explicitly between pension and welfare benefit plans in § 514(a) is understandable, for, as explained above, the managed care plans that wreak havoc with § 514(a) as it relates to welfare benefit plans did not exist when ERISA was enacted. There is no evidence that Congress envisioned the current situation.

Taking note of Congress's understandable lack of clairvoyance, the Supreme Court might embrace pragmatism and limit § 514(a)'s preemptive scope regarding welfare benefit plans. Although the distinction would find little support in the text of ERISA itself, Justices Scalia and Ginsburg recently admitted in a concurrence that:

[A]pplying the `relate to' provision according to its terms was a project doomed to failure, since, as many a curbstone philosopher has observed, everything is related to everything else. The statutory text provides an illusory test, unless the Court is willing to decree a degree of pre-emption that no sensible person could have intended — which it is not.

California Division of Labor Standards Enforcement v. Dillingham Construction, N.A., Inc., 519 U.S. at 335-36, 117 S. Ct. 832 (1997) (internal citation omitted). Absent textual guidance or meaningful legislative history, there is little to prevent a pragmatic solution to a problem Congress has not confronted.

No doubt there are other possible solutions. The vital thing, however, is that either Congress or the Court act quickly, because the current situation is plainly untenable. Lower courts are routinely forced to dismiss entirely justified complaints by plan participants who have been grievously injured by HMOs and plan sponsors, all because of ERISA, the very purpose of which was to safeguard those very participants. Our dockets grow increasingly crowded with cases where participants offer myriad varieties of artful pleadings in their desperate attempts to circumvent ERISA's procrustean reach, and our caselaw grows massively inconsistent due to the sheer complexities of the subject and lack of any meaningful guidance. There must be a better way.

The Clerk of Court is directed to send a copy of this opinion (with attention directed to the concurrence) to the Solicitor of the Department of Labor; the Chair, Ranking Member, Chief Majority Counsel, and Minority Counsel of the Senate Committee on Health, Education, Labor, and Pensions; and the Chair, Ranking Member, Chief Majority Counsel, and Minority Counsel of the House Committee on Education and the Workforce.

AMBRO, Circuit Judge, concurring.

Judge Rendell's opinion is well-crafted, and I join it. In many cases, however, the result underwhelms. Thus, like Judge Becker, I implore for a better way to make these kinds of decisions.

Congress in 1974 passed "a comprehensive statute designed to promote the interests of employees and their beneficiaries in employee benefit plans." Shaw v. Delta Air Lines, Inc. 463 U.S. 85, 90, 103 S. Ct. 2890, 77 L. Ed. 2d 490 (1983). That statute — the Employee Retirement Income Security Act of 1974 ("ERISA"), 29 U.S.C. § 1001 et seq. — made policy compromises in order to combat exponentially rising health care costs in a melange of state tort laws assessing negligence after-the-fact. See generally Jonathan J. Frankel, Medical Malpractice Law and Health Care Cost Containment: Lessons for Reformers from the Clash of Cultures, 103 Yale L.J. 1297 (1994). Among the newer forms of managing the providing of healthcare services was to "`prospectively and concurrently assess the appropriateness of care....'" Id. at 1302 (quoting Elizabeth W. Hoy et al., Change and Growth in Managed Care, HEALTH AFF., Winter 1991, at 27).

Instead of healthcare insurers reimbursing insureds for already-given medical care, health maintenance organizations ("HMOs") were created where, upon "receipt of a fixed fee for each patient enrolled under the terms of a contract [,] ... specified health care [is provided] if needed." Pegram v. Herdrich, 530 U.S. 211, 218, 120 S. Ct. 2143, 147 L. Ed. 2d 164 (2000). Physicians contract with HMOs to provide services to patients in the HMO for set fees. HMOs, via a health plan purchased by patients' employers, set "[r]ules governing collection of premiums, definition of benefits, submission of claims, and resolution of disagreements over entitlement to services...." Id. at 223, 120 S. Ct. 2143. The bottom line — non-routine medical services are provided after they are approved.

This system, simple in concept, has sisyphean frustration in application. At the forefront is preemption. To save costs in a tort system where medical decisions by service providers are second-guessed by non-medical persons, ERISA preempts state law that conflicts with ERISA, 29 U.S.C. § 1144(a), and is a "cause of action within the scope of the civil enforcement provisions of [ERISA] § 502(a) [29 U.S.C. § 1132(a)]." Cicio v. Does, 321 F.3d 83, 94 (2d Cir.2003) (quoting from Metropolitan Life Ins. Co. v. Taylor, 481 U.S. 58, 66, 107 S. Ct. 1542, 95 L. Ed. 2d 55 (1987)). Included within this subsection is any civil action brought "by a [plan] participant or beneficiary ... 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...." 29 U.S.C. § 1132(a)(1)(B). Stated colloquially, ERISA "federalizes" most employee benefit plan remedies.

Judges Rendell and Becker in this case, and Judge Calabresi in his partial dissent in Cicio, 32 F.3d at 106 et seq., point out poignantly the so-often unsatisfying constructs we use to cipher meaning when cultures of cost containment and tort responsibility clash over preemption. "The litigants in these cases spar over the extent to which a lay financing entity [the HMO] has appropriated to itself distinctly medical authority by substituting its own expertise for that of the treating physician." Frankel, 103 Yale L.J. at 1303 (emphasis in text). I join the chorus calling for a fresh look from higher authority to promote better the primary Congressional purpose of ERISA — protecting plan participants and their beneficiaries. "At issue is nothing less than the definition of medical decision-making itself," id., and the lives that this affects.