Evidence-Based Medicine and the Law: The Courts Confront Clinical Practice
Basic Concept and Operation of the Certification Program
Stated most simply, the proposal described more fully below calls for the establishment of a federal certification program for clinical practice guidelines, operated out of the Department of Health and Human Services (HHS), presumably by the Agency for Healthcare Research and Quality (AHRQ). The agency running the program would have responsibility for vetting, certifying, supervising periodic updating, and helping to promote the dissemination and adoption of CPGs. The program would be voluntary and open to any person or entity wishing to put a guideline, or set of guidelines, forward for certification. Basic, fixed costs of running the program would be funded through the federal budget; variable costs would be covered through fees charged to those submitting guidelines for review. The process of certification would involve not just government personnel but also representatives of the provider and consumer communities. Certification, the federal "seal of approval," would give a guideline greater credibility, making it more "saleable" in the health care marketplace and would also entitle it to special treatment in federal and/or state courts in litigation involving quality-of-care and entitlement-to-benefits, or coverage, claims. Essentially, a certified guideline would be accepted by the courts as setting the legal standard for acceptable quality of care in a malpractice suit, or for determining in a coverage dispute whether a given treatment was medically appropriate. Since fixing the applicable standard of care is a core issue in most malpractice and coverage litigation, providing a simpler way to resolve this issue would significantly streamline the litigation process, saving time and money for all involved.
Who Should Do the Certifying?
Certification of guidelines could be done on the private or public level and, if on the latter, by either a state or a federal agency. The last of these alternatives clearly seems to be the best. Accreditation programs run by nongovernmental entities—such as the National Commission on Quality Assurance (NCQA), which accredits managed care organizations, and the Joint Commission on the Accreditation of Healthcare Organizations (JCAHO), which accredits hospitals and other institutional providers—have been used as adjuncts to governmental regulation for many years and have a commendable track record. However, although the success of entities such as NCQA and JCAHO shows that private organizations can contribute importantly to the definition and safeguarding of the quality of care, because the proposed CPG program contemplates application of its output in the courts, it is simpler and more certain to have a single, governmental agency setting and implementing the standards for certification. The main purpose of certifying guidelines is to help courts distinguish those that are valid and credible from those that are of questionable merit. Having a single, official certification process best supports this objective; having multiple certification processes and standards would reintroduce complexity and uncertainty into the situation and would thus be counterproductive.
Assuming, then, that a government certification program is the preferred approach, should the process be handled at the state or federal level? Some have advocated state programs over a federal program, at least as an initial step (Trail and Allen 1995-96: 250). However, even though a few states, such as Maine, have taken on the burdens of certifying guidelines (24 Maine Revised Statutes Annotated, Section 2973), the task ultimately is too large and complex to be administered effectively at the state level. To handle not just the initial certification of guidelines but also periodic updating to recognize and incorporate advances in scientific knowledge and medical practice, there would have to be a highly skilled core staff for whom review of guidelines was their main activity. Maintaining consistently high-quality certifying agencies at the level of the several states would be problematic. Furthermore, the waste and redundancy inherent in duplicating certification activities in the states, increasing cost not just for the states but also for guidelines issuers and users, many of which are national in scope, makes a state-level process highly undesirable. Finally, with a state-level process would come the possibility that a guideline certified in one state might be disapproved in another, opening the door to the very kind of sorting-out confusion the certification program is intended to avoid. Despite the obvious negatives to state-level certification, states would still be allowed to engage in this activity if they so chose; in fact, it is hard to conceive how they could be barred from doing so. It is unlikely, however, that many states would choose to undertake this complex, labor-intensive, and expensive activity if a federal program were in place. If some states did, they would at least have the federal model to look to for guidance.
Assuming that a federal-level program is the way to go, an existing entity, AHRQ, a unit within the HHS, is the obvious choice to perform the guideline-certifying function. This choice acknowledges AHRQ's core mission and its substantial contributions over more than a decade—largely under its former name, the Agency for Health Care Policy and Research (AHCPR)—toward developing and refining the methodology for generating CPGs. It also recognizes AHRQ's recent efforts, with ECRI (known in the 1960s as the Emergency Care Research Institute), a nonprofit healthcare research organization based in suburban Philadelphia, to compile and maintain an extensive on-line database of practice guidelines, the National Guidelines Clearinghouse™ (NGC).3 In fact, a Web-accessible database like the NGC, and the organizational infrastructure to support it, is a sine qua non for a workable certification program, further confirming the appropriateness of looking to AHRQ.
It should be clearly understood that, while AHRQ would provide leadership and technical assistance in the development of guidelines, and would work closely with CPG developers, it would not develop guidelines on its own. Private-sector activities supporting the development of CPGs should be encouraged and relied upon to the fullest extent practicable. Moreover, nongovernmental parties representing professional interests and the health care-consuming public should be involved in the review of guidelines. The government's role in defining standards for medical practice should not extend beyond what is necessary to assure the safe and sensible development and application of CPGs.
Should Certification Be Mandatory or Voluntary?
The guidelines certification program should be voluntary. In fact, it is hard to imagine how a mandatory program would work. The FDA (Food and Drug Administration) process for reviewing the safety and efficacy of drugs is mandatory, in that no manufacturer is allowed to market drugs that have not been approved. But given the pervasiveness of guidelines and the infinite number of forms they can take, it would hardly be possible to say that no guidelines can be used unless approved by the government. In a voluntary program, the decision to seek certification of a CPG would rest entirely with its issuer, as is presently the case when a guidelines issuer elects to submit its CPG for inclusion in the NGC database. The incentive to seek certification would be great, however, as there would be benefits to having one's CPG certified, especially with regard to the way certified guidelines are treated in the courts. As discussed below, courts would be directed by legislation to treat certified CPGs as setting the applicable standard of care in cases involving quality of care or coverage under health plans.
The Criteria for Certification
To be certified, a CPG would have to be developed: (1) through rigorous, scientific outcomes research, based upon an appropriate and adequately large set of clinical practice data; (2) using appropriate methodology, as defined by AHRQ regulations; (3) with input from qualified medical professionals; and (4) with provision for frequent updating, to incorporate advances in medical knowledge generally as well as experience gained through clinical practice using that particular CPG. Guidelines developers would be required to maintain an efficient system for the ongoing monitoring and evaluation of experience with the clinical use of their guidelines. This would be similar, in concept at least, to the mechanism by which the FDA requires pharmaceutical manufacturers to track clinical experience with their newly approved drugs.
A further requirement for certification would be that the guidelines must be well written; they must be sufficiently clear and easy to follow that clinicians and others seeking to use them can do so confidently and safely. Just as there is a science to developing the content of CPGs, there is an art to expressing that content coherently. Here the expertise of AHRQ would necessarily be brought to bear; a guideline that the reviewers could not understand would hardly pass muster.
The Significance of Certification
Certification of a guideline would offer a number of benefits to its developer and to those who used it. First, the credibility bestowed by the governmental imprimatur, an assurance that the guideline was properly developed and well expressed, would assist the developer in getting people to adopt it. If the guideline were being developed for sale—for example, by an outcomes research organization that used EBM methodologies to write cost-containing guidelines for use by MCOs—the developer could command a better price for a certified guideline, in large part because of the way courts would treat a certified CPG as compared with one not certified. A certified guideline would be admissible in court as evidence of the appropriate standard of care without further proof of validity. Moreover, compliance with a certified guideline would raise a presumption of proper care.
To offer a concrete example, suppose an HMO were to disseminate to its participating physicians a guideline that called on them to use more sparingly an expensive test that previously had been in routine and frequent use. Assume that this cost-reducing guideline was based on recently completed outcomes research that showed that conventional, more liberal use of this test wasted resources without yielding a discernible benefit in terms of patient outcomes. The HMO could gain a significant cost-competitive advantage by promoting less frequent use of the test, but its participating physicians might well be concerned that if one of them happened to experience a bad result apparently associated with departure from standard practice, the physician's deviation from the professional community standard of practice would expose him or her to malpractice liability. The court's direction to regard compliance with a certified guideline as proof—either conclusive or rebuttable proof that the care rendered was adequate and thus not tortious—would be a powerful counterforce to the physician's reluctance, on legal grounds, to depart from traditional overuse of the test. This hypothetical example reflects, by the way, a fundamental, recognized purpose of CPGs, to help overcome the resource-wasting phenomenon of "defensive practice of medicine" (Daly 1995).
As to whether compliance with a certified guideline should raise a conclusive or a rebuttable presumption of adequate care, arguments can be made for either position. A conclusive presumption would effectively give certified guidelines the force of law and would be the simpler approach. Proof of compliance—or proof of noncompliance if the guideline were being used by the plaintiff for inculpatory, or "sword," purposes—would end the inquiry as to the adequacy of the care given. This approach seems too pat, however; and cases could arise in which justice would demand further consideration. For example, suppose a patient suffered harm when her physician followed a certified guideline that, it could be proven, he knew was rendered arguably obsolete by recent scientific findings but that he chose to follow anyway. (Assume, for example, that a new guideline had been certified, as the defendant knew, but the old guideline had not yet been retired.) Allowing compliance with the older certified guideline to shield the defendant's actions completely would tend to chill the advance of medical knowledge, an acknowledged danger of certification processes generally. Rebuttal of the presumption of adequate care should be allowed, but should not be too easy, or else it would undermine excessively the value of certification. A reasonable middle position—perhaps along the lines of Senator Chafee's 1993 health care reform proposal mentioned above—would be to let the presumption be rebutted but only by "clear and convincing evidence." This recognized standard of proof lies between the traditional civil standard of "preponderance of the evidence" and the much stricter criminal standard of proof "beyond a reasonable doubt."
Unlike the Maine and Minnesota guidelines programs described above, the proposed program would treat evenhandedly the parties to litigation implicating CPGs. Plaintiffs should be allowed to use certified CPGs on the same basis as defendants, eschewing the Maine/Minnesota "shield—only" or "one-way street" approach. While CPGs may be more useful to one party than the other in some circumstances, access to their use and the basic procedural and evidentiary rules should be the same for both. One important reservation to this principle should be noted, however: the presumption of proper care should prevail. Where two (or more) certified guidelines were in conflict, with the plaintiff asserting noncompliance with one ("sword" use) and the defendant relying upon compliance with the other ("shield" use), the presumption that the defendant's care was adequate should be upheld. If the defendant health care provider could show compliance with an appropriate, certified guideline, he or she would be entitled to the presumption that the treatment rendered was adequate to meet the legal standard of care, even in the face of a contrary, certified guideline proffered by the plaintiff-patient. To apply this to the breast cancer screening example above, if both the National Cancer Institute's guideline calling for biennial breast cancer screening for women in a given age and risk category and the American Cancer Society's guideline recommending annual screening for this same population were certified, the provider, the HMO in this case, would prevail. Such treatment would be the functional equivalent of the "respectable minority" doctrine currently applied in many U.S. jurisdictions.
Note that, while this approach would greatly simplify the court's job of determining what standard of care to apply and would give the defendant provider an important "benefit of the doubt," it would neither answer all questions nor assure the defendant of victory. In many cases, an important issue might remain as to whether the guideline(s) proffered was (were) appropriate to the facts at hand. No matter how much benefit of the doubt was given to a particular CPG, it would not carry the day unless it was found to be an appropriate guideline for the situation at bar. Likewise, there might be substantial factual dispute as to whether the guideline was, in fact, followed by the defendant and, if it was not, whether the harm suffered by the plaintiff was caused by the defendant's noncompliance. The only thing changed by the guidelines certification program would be the simplifying of the court's search for the proper standard of care to apply; but that one change would be a most significant contribution to the cause of judicial efficiency.
What treatment should be accorded to an uncertified guideline under this regime? This is a tricky question that deserves closer consideration and debate before a firm position is taken. As a starting point, it seems clear enough that a court should be able, in its discretion, to admit an uncertified guideline into evidence if satisfied that the guideline meets the criteria for validity specified in the certification program. Presumably, courts would generally be disinclined to engage in such an inquiry; and issuers and users of guidelines would be motivated to undertake certification to avoid the uncertainty of not knowing how a given court would react to their uncertified guideline. On the further question of what weight and treatment to give certified as opposed to uncertified guidelines, things are less clear. Putting uncertified guidelines on substantial parity with certified ones would undermine the incentive to seek certification; however, discounting their evidentiary value too greatly would belie the claim that the certification program is voluntary. Numerous lesser questions can be raised. For example, if an uncertified guideline is admitted, should the court allow inquiry into why it is uncertified? Is it because the developer did not seek certification, or was it sought and denied? Should the court allow the jury to be led through an exploration of why it was denied? These are subtle questions indeed. In addressing them, policy makers should keep in mind that a main purpose of promoting the use of guidelines in the courts is to simplify and rationalize the litigation process, not to add new and uncharted dimensions of complexity.
One final, but very significant, point should be noted about the implications of certification. The developer and/or issuer of a CPG should be immune from civil liability for harms caused by its adoption and use if the requirements for certification have been met. This would seem to follow logically and implicitly from the foundation principle that compliance with a certified guideline shields the user of the guideline; but the implication for the guideline developer/issuer should be explicitly acknowledged. Although it is unclear whether guidelines issuers might be liable under current law, there has been scholarly speculation on the issue (see e.g., Brennan 1991: 78-81; Rosoff 1995: 91-93), and the possibility is real enough to provide additional motivation to issuers to seek certification.
The Costs of a Certification Program
The proposal is simple to state on this point: general costs related to AHRQ's activities as the certifying agency should be borne by the federal budget. Reasonable costs of initial certification, and of periodic recertification, should be borne by the entity issuing the CPG. This approach roughly follows that used by the FDA in its approval process for drugs and medical devices; and there is ample precedent to be found in other federal regulatory programs as well. A more detailed analysis of the likely costs of a federal guidelines certification program with both the on-budget and off-budget costs is beyond the scope of this essay but would, of course, be an important consideration in any legislative debate about adopting such a program.
The Challenges of Implementation
The guidelines certification program proposed here poses a number of challenging issues, not the least of which is the daunting task of gaining sufficient support within the administration and in the Congress to have it put forward and enacted. Without meaning to minimize that concern, the simple answer is that it is beyond the scope of this essay. In any event, unless some consensus can be developed that a certification program of the type proposed is worth having, there is little point in strategizing about how to get it passed.
Assuming adequate political support for a certification program of the type envisioned, a central question is how its proposed impact on state law and state court processes would be achieved. One would hope, of course, that states would see the value of the certification program and would want to support it through their court systems without any compulsion. If this were not the case, at least with particular states, other means of attaining state participation would have to be sought. This is certainly not a trivial problem. While a full treatment of this issue would extend the article well beyond the constraints of this symposium, it would be cavalier not to address it, at a basic level at least.
Both the provision of health care and the operation of the insured and employer-self-funded plans that pay for much of it are so intertwined with interstate commerce that Congress likely has the jurisdiction to regulate it under the commerce power of Article 1, Section 8, of the Constitution. Regulation of health care-related litigation at the state level, however, implicates state sovereignty concerns in a way that regulation of health care alone does not. ERISA, the Employee Retirement Income Security Act of 1974, provides a highly visible precedent for federal preemption of state law in this area, although, in the view of many commentators, it is not a particularly positive precedent (see, e.g., Mariner 1996, 2000). If implementation of the state law aspects under the federal Commerce Power is deemed too problematic, there is also the possibility of using the Spending Power, also under Article 1, Section 8. States could be offered financial incentives for voluntarily participating in the program, much as federal highway funding was tied in the 1970s to states' adoption and enforcement of the 55-mile-per-hour speed limit. Finally, the certification program could be tied to other federal health care programs, principally Medicare and Medicaid, loosely following the precedent established by the Emergency Medical Treatment and Active Labor Act of 1985 (EMTALA), which requires hospitals participating in Medicare to provide urgent care to all of their patients, not just those with Medicare coverage. There are numerous other examples of federal health care initiatives being successfully implemented at the state level either through the Commerce Power (e.g., the 1974 Health Care Amendments to the National Labor Relations Act, Pub. Law 93-360, codified in Section 152 of Chapter 42 of the U.S. Code) or an artful use of "carrot and stick" mechanisms (e.g., the National Health Planning and Resources Development Act of 1975, Pub. Law 93-641, 42 U.S. Code, Section 300K et seq.). Given this range of possibilities and precedents, it is fairly certain that the jurisdictional premises upon which the program rests can be successfully handled.
Assuming the program can be enacted, an important ongoing concern is that the certification process not be allowed to freeze, or even chill, the advance of medical knowledge and practice. This issue has been touched on above and is not easily resolved. The program, as implemented, must provide a prompt, efficient, and low-cost process not just for the review of new guidelines but also for the timely updating and revision of approved guidelines, including the decertification of guidelines known to be no longer valid. There must be adequate budget to assure an ample number of competent reviewers and an ongoing commitment to rapid turnaround of materials submitted. The use of electronic submission of materials and subsequent communication about them should be a significant help in this regard.
Proposals for federal certification of clinical practice guidelines, including not only my own freestanding proposal but also those contained in the Clinton, Chafee, and other federal health care reform proposals, have been "around" for over five years without attracting substantial attention or support. This is true despite the growing acceptance and use of CPGs during this period. If anything, the continuing proliferation of CPGs makes it more advisable that some mechanism be devised to introduce guidelines more prominently into the legal process and help courts decide which guidelines should be regarded as authoritative. I hope this present restatement and more detailed reexamination of my proposal in the context of this larger consideration of EBM will help to raise awareness that the time for a national guidelines certification program has come—or, at least, is on its way.
If the above general principles and specific elements can be fashioned into a national framework for the certification of clinical practice guidelines and a rationalized approach to their use in the courts, I believe the expanded use of CPGs it will enable will bring substantial benefits to our health care system. While there will always be a need and a place for professional medical judgment, we should make maximum use of available empirical evidence as to what works and doesn't work, synthesizing that data into carefully analyzed, widely disseminated guidelines to assist physicians to properly apply their clinical judgment. Benefits will flow to the legal system as well, making possible more accurate, efficient, and affordable resolution of disputes about the quality and appropriateness of health care provided.
References and Notes
Lewin, L., and J. E. Erickson. 1989. Leadership in the Development of Practice Guidelines: The Role of the Federal Government and Others. Prepared for the PPRC's Conference on Practice Guidelines, Washington, DC, 11 October 1988, revised 24 April 1989.
1. Mulrow and Lohr cite Sackett et al. 1997, as the source of this definition.
2. For example, where the issue is causation, the question might be put as follows: "Doctor, in your expert opinion, based upon your knowledge and personal experience, do you believe that the defendant's acts (or omissions) caused the plaintiff's injury?" Similar examples could be offered involving assessment of damages ["Doctor, how serious is the patient's injury? In your view, is this a case of "total disability?"] and prognosis ["Doctor, in your opinion, will the patient-plaintiff ever again be able to drive a truck?].
3. For further information on NGC activity and a firsthand look at its work product, visit http://www.guideline.gov. For background on the ECRI, visit www.ecri.org.