Our goal is to offer, by way of a test case, tangible recommendations for accountability. We focus on a domain that is often ignored in discussions of accountability, namely, the ethical dimensions of medical care—practices, such as informed consent, that are distinct from diagnostic and procedural skills. Of all the possible domains of accountability, we concentrate on ethics because it is among the most challenging to evaluate and because it is crucial if the medical profession is to maintain the trust of patients and the public [5, 6]. In the absence of a national body, the Working Group on Accountability convened to identify the norms of ethical medical practice and to delineate desirable methods of assessment. It is an ad hoc working group of clinician ethicists; most members and consultants are practicing clinicians, and each has a specialty focus in professional ethics, medical assessment, organization, or policy.

We start by summarizing a guiding model of stratified accountability, which is described in the accompanying article [7], and we then develop the model in the specific domain of ethical practice. We conclude by suggesting specific initiatives with which to implement the recommendations.

A Guiding Model of Accountability

Accountability consists of three main parts: the loci or accountable parties, the domains or subject matter for which these parties are accountable, and the procedures by which these parties are assessed and held accountable. Modern medicine consists of many accountable parties, and each can hold the other accountable over many different domains; this leads to what might be called a complex reciprocating matrix of accountability. For example, physicians are accountable to patients, hospitals, employers, the government, and others, and in turn they can hold hospitals, insurers, the government, and others accountable. Models of accountability differ in the domains that are considered to be primary, the specific content areas within each domain that are to be assessed, and the procedures by which different parties will be held accountable. In the accompanying article [7], we argue that no single model of accountability is appropriate to the complexities of modern medicine. Instead, we endorse a stratified model of accountability. At the core is the traditional professional model, which applies to individual physician-patient interactions and to some institutional matters. Surrounding this core are other institutional matters and interactions among health care institutions in which political and economic mechanisms of accountability, such as public forums to evaluate and approve policies and financial accountability, seem appropriate.

In such a stratified model, physicians have primary responsibility for characterizing the content areas in which they will be held accountable, the mechanisms for assessment, and the processes for explanation and action. In this view, professional values remain central but are not sovereign; through the political and economic models of accountability, patients, employers, and others participate in approving and evaluating domains and mechanisms of accountability. By using medical values along with structures of continuously adjustable and testable fairness, the model can combat both public cynicism about "medicine's self-interest" and the profession's annoyance with "ill-informed intrusive outsiders."

Thus, in the next three sections, we begin the professionals' process of delineating the core content areas, the assessment procedures, and the mechanisms for response for medical ethics. We identify eight content areas for accountability in medical ethics; because this list is for accountability mechanisms, each area is subdivided into assessable items, numbered to match items in Table 1. Two principles inform this list. First, the content areas are culled from traditional codes [8], from recent statements by medical bodies [9, 10], and from consensus among the members of the Working Group on Accountability. Each area can be derived from fundamental ethical principles; references are provided. Our selection criteria, in keeping with the stratified model of accountability, were to cover professionally endorsed medical values for which patients and other parties may legitimately request accountability. Second, this list is necessary but may not be sufficient; it may be modified as the deliberative process of accountability evolves. We do not propose that professionals establish a rigid, unchanging list to which they hold themselves accountable or that nonprofessionals blindly accept and use such a list. Rather, we propose that professionals create a list that other groups can modify and that professionals and other groups hold themselves and one another to these deliberated standards.

Core Content Areas of Accountability in Medical Ethics

Medical Decision Making

Medical decision making is a pre-eminent area of desirable accountability in medical ethics. Patients have well-recognized rights to be informed of their health status and to share in decisions about their care [11-14]. For assessing physicians, we propose that medical decisions be considered in the following contexts of care: noninvasive tests or therapy decisions (item Ia, Table 1); invasive tests or intervention decisions (Ib); life-sustaining intervention decisions (Ic); decisions on participation in clinical research (Id); and decisions in any of the above categories for patients not competent to make their own decisions (Ie). The Joint Commission on the Accreditation of Healthcare Organizations has articulated counterpart standards for institutions [15].

Confidentiality

Another well-recognized duty of the medical profession is to maintain patient confidentiality to the greatest possible degree, yielding only to the most pressing and direct interests of others, such as in cases of threats to life [8-10, 16]. Physicians should be accountable for their assessed knowledge of and adherence to standards of confidentiality (IIa). Institutional record-keeping practice is especially vulnerable in the current era of electronic records. Use of medical data for utilization review and patient targeting, such as screening-test eligibility, is widespread, and the sharing of information in integrated health networks may bring further possibilities for breaches of confidentiality [17]. Institutions should therefore be specifically accountable for the following: precautions taken to protect confidentiality (IIb); the policy for release of person-identified information—what, to whom, under what circumstances, and when it is legally required (IIc); the regulations for use of medical record data for research, both private and governmental (IId); and the type of information accessible to third-party payers (IIe).

Fiduciary Obligations and Conflicts of Interest

Physicians and institutions deal with many forces that may compromise the fundamental mission of patient care [18-28]. That physicians must act for the patient's good in preference to their own is widely accepted [29, 30]. For accountability, it is possible to focus on five types of approaches to limit the negative effect of conflicting interests. The following three approaches are for physicians and institutions: documented education of physicians and managers about conflicting interests (IIIa); required disclosure to the institution or patient and institutional review of specified types of conflicts by physicians and managers (IIIb); and physician or institutional disclosure with patient informed consent whenever appropriate (IIIc). Two further approaches apply to institutions: prohibition of physician conflicts that undermine the fiduciary relationship (IIId); and policy safeguards that limit conflicting incentives for institutions (IIIe).

Responsibilities Arising from Patient Vulnerability

Much of medical ethics rests on the injunction to avoid exploitation of patient vulnerability, but some matters deserve particular concern. Serious boundary violations, such as sexual relations with patients, risk egregious patient harm. Physicians should be clear about the basis for the profession's categorical ban on sexual relations with patients (IVa) [31, 32]. Awareness of sometimes less obvious violations, such as inappropriate touching, gift giving, and socializing, can be helpful toward this end. Because boundary violations are difficult to discuss, institutions should ensure that ethical standards on the matter are explicitly expressed during physician training and continuing education (IVb) [33]; they should also have a policy statement, which should be publicly available and readily accessible through patient representatives, ombudspersons, or other patient resources (IVc).

Patient vulnerability can also be an issue in clinical research. The potential for serious personal damage resulting from research exploitation necessitates accountability on these matters. Thus, physicians must follow informed consent procedures and other regulations about human subjects research (IVD), training should be mandatory, and regulations should be publicly available [34].

Practitioners' Personal Standards

Physicians' personal conduct becomes a matter of accountability when it affects patient care. Physicians and trainees should inform patients about their training or expertise level when they introduce themselves (Va) [35, 36], seek care for problems with chemical dependence or other illnesses or problems that could put patients at significant risk, and submit themselves to mechanisms that ensure patient protection (Vb). Institutions must promptly address serious risk to or actual compromise of competent care, usually by suspending care privileges. Matters that are not expected to affect patient care, such as the care of patients with noncommunicable, nondebilitating illness, can be addressed through less public means. Policies for evaluating and dealing with private matters should, however, be available to requesting parties (Vc).

Equity among Patients

Because of the complex nature of the subject, the just allocation of health care resources is usefully considered under a separate domain. Nonetheless, because it is clearly a matter of ethics, aspects relevant to physicians and institutions are included here. Individual physicians, governing boards, and directors of health care facilities have an obligation to consider all patients with medical needs that fall under their areas of practice or coverage. Under our model of accountability, these health care professionals are therefore obliged to reveal how much they do to ensure fair access to care for all patients (VIa) [37-41]. Individual persons (VIa1) and organizations (VIa2) all face some resource limits and therefore routinely make allocation decisions. Such decisions include triage at the medical care site (such as emergency wards, intensive care units, and dialysis units) (VIb1) and personnel, space, and budget allocations among areas in medical care (such as primary care services and intensive care units) (VIb2). Publication of statistics and institutional policy or guidelines on access and on triage and allocation decisions would provide processes of accountability.

Cultural Representation

Because health is a prerequisite for so many other goods in life, medicine should seek to ensure that health care is not compromised by discrimination on the basis of culture, race, sex, sexual orientation, or other identity differences. The finding that care continues to differ along these lines justifies certain corrective mechanisms [42, 43]. Appropriate corrective resources may include training for physicians about the relevant groups (VIIa), translation services (VIIb), and providers from the groups served (VIIc).

Procedures for Resolving Dilemmas and Expert Advice

Recipients of health care can reasonably expect that mechanisms exist for resolving common ethical issues, such as disagreements over withdrawing life support or access to limited medical resources [15]. Recommendations for due process in ethical consultation and ethics committees are not yet at the level of professional guidelines but are evolving rapidly [43-47]. Patients should be able to known about the availability of expertise in ethics at the hospital, including consultants in medical ethics (VIIIa), ethics committees (VIIIb), consultants or committees accessible to patients (VIIIc), and patient representatives or ombudspersons (VIIId).

Procedures for Assessing Conduct, Standards, and Policy

It is a serious matter when accountability is asserted by one party over another, and methods of assessment must be good. There should be a priori agreement that methods of assessment for accountability be as objective and as rigorous as possible. This applies to each assessment area and to comparisons among individual persons or institutions. Thus, for example, methods must be standardized when comparisons are being made. To assist in validating and standardizing assessment mechanisms for medical ethical practice, we offer the following perspective and suggestions.

As with more technical aspects of medical practice, measurement of ethical aspects of medical practice is not simple [48, 49]. Formally validated questions and graded scales to assess ethically relevant conduct remain underdeveloped. However, there is no inherent barrier to assessing ethical aspects of practice. Some related efforts, such as studies on appropriate and inappropriate care, have already been proven important [50]. Other assessment instruments that use a range of information, such as the Healthcare Employer Data Information Set, are already in widespread use. Such approaches in the domain of ethical conduct should be able to achieve greater accuracy than existing mechanisms that rely on anecdote, general reputation, or colleague reports. As the science of assessment develops, measures may change. Indeed, a critical force in the evolution of accountability will be the assessment of assessment sciences themselves [51-56]. Nevertheless, assessment methods currently are sufficient to allow the profession to start making good use of them and, more importantly, to help develop the discipline, in this case as it applies to the ethical domain in medicine.

For ethical practice and using the stratified model's requirement that assessment be by and for multiple parties, the following methods for assessment are currently likely to be most useful: use of periodic public, patient, patient family, and provider surveys (or other methods for gathering feedback) about physicians and institutions; formal accreditation methods, which should include peer review; possible limited use of record audits and reviews; provision of publicly available institutional policies and procedures; and periodic publication of report cards that provide a composite of results from the preceding list of assessments. These methods are selected for their inclusion of all involved groups and for their amenability to methodologic rigor. As newer methods of assessment develop, they may supplement or even replace these mechanisms.

Surveys

Surveys of providers, patients, and the public, already used for much of the empiric research in medical ethics, can provide perspectives on the process and outcome of practices. Table 2 shows how physicians and institutions may be evaluated by patient survey for the eight content areas in medical ethics.

Although survey methods have important limitations, they can measure two kinds of ethical practices. First, surveys can measure satisfaction with or personal perspectives on the facilities and processes of care [57, 58]. This aspect of survey methods is relatively well developed, and validated instruments for patients are already available for some aspects of care [59]. Although they may overlap, satisfaction and quality of care are not the same; satisfaction should therefore be used as only one of many measures [60, 61]. Second, surveys can describe objective accounts that may be interpreted as indicators of quality of ethical conduct, such as engagement in an informed consent procedure, availability of a chaperon during a genital examination, or availability of a translator when a language barrier is present. The work of several investigators can provide models of how the instruments might be adapted and designed for ethics [62, 63].

Surveys can also serve as more than a procedure of evaluation. They can provide suggestions on, for instance, the need for new content areas. By including in surveys a section for suggestions, unexpected perspectives can be articulated and incorporated. Furthermore, in the design of validated survey instruments, good methods require consultation with all relevant parties in, for example, focus groups. Such focus groups may suggest additional areas of evaluation or specific questions. Surveys thus can become an essential part of the process by which nonprofessionals review and evaluate content areas and assessment methods.

Accreditation

We endorse the practice of including ethics in accreditation procedures, both for physicians at certification and licensure and for institutions. Relatively sophisticated psychometric methods are being used for testing in Board assessment procedures, and increasing attention is being given to ethics [64]. Because health care recipients should have a say in the standards by which professionals are judged, we endorse the view that accreditation bodies should encourage lay representation in standard setting procedures.

Record Review

Existing forms of medical record review are likely to provide inaccurate measures of individual physicians' standards because ethical conduct tends to be undocumented. These methods should therefore be used only in combination with other methods and with suitably cautious interpretation [51, 52, 65]. Summary statistics, however, may provide a reasonable measure of some aspects of ethical practices, such as the presence of informed consent forms or the match of actual care with overall patient preferences [66]. Novel types of routine review are also possible, such as use of the data banks of Medicare, Medicaid, or other large providers to track an organization's provision of care to vulnerable populations. Nevertheless, such methods can only be recommended for development and are not yet available for widespread use.

Institutional Policies and Procedures

Knowledge of the institution's policies and procedures is an indirect form of assessment. For example, knowledge that the requirements of the Joint Commission on Accreditation of Healthcare Organizations are rigorous provides some assurance to patients. Similarly, the existence of consultative services and committee functions for ethical dilemmas allows patients a way to assess the attention paid to appropriate processes. Policies on physicians' private matters, such as health, similarly allow patients to assess institutional attention to the responsibility of their providers.

Rendering Account and Taking Responsibility

Once assessment has been made in agreed-on areas, the assessed party must be called to account; that is, be allowed to justify actions and required to take appropriate responsibility. Accountability allows reassessment in light of the responding account and then requires action by the accountable party to improve performance and compensate for shortfalls. This implementation stage is difficult and often imperfect. Nonetheless, it is necessary and is currently proceeding without much professional recommendation to guide it. We offer the following tools to tie our professionally recommended standards and measurements to an implementation structure that responds to the multiple perspectives required by the political and economic models of accountability.

Report Cards

We endorse the use of composite assessments as a way to provide information to otherwise poorly informed persons who need to ask pertinent questions of the accountable parties. Thus, not only patients and health plan purchasers but also health economists, health policy experts, and political leaders or candidates can access information in a consumer report format. To report on standards in ethical aspects of medical practice, a compilation of results from the identified assessment methods in each of the itemized content areas is possible. We outline a prototype "report card on ethics" for assessing physicians and institutions (Table 1). In a good system, publication of report cards should be delayed to allow for due process in accountability. Furthermore, in the full system of stratified accountability, relevant information on the other parties—patients, medical educators, health policy experts, and so forth—would also be available in an expanded report format. Finally, in a responsive, multilateral community of accountable parties, it should be possible to comment on the desirability of the content areas included in or excluded from a report card.

Exchanging Perspectives

The accountable party should be able to receive all relevant information from the assessments. Channels for communicating assessment information to physicians and institutions can include personal letters, open letters from organization managers, presentation of summary statistics during mortality and morbidity rounds, circulation of institutional report cards, and publications in peer journals. A combination of public and provider opinions, which was used in Canada's Pickering Report on the state of the profession [67-71], could be used in internal institutional processes and open professional publications. Responses from the accountable party to the accounting parties can use similar channels. Patient-owned health care organizations, such as the Group Health Cooperative of Puget Sound, are naturally compatible with communication mechanisms between most of the directly involved parties.

Public deliberation is a seldom-used mechanism in the medical field but one that has great potential. Small "town meetings" can be a useful forum. On a larger scale, the nature and extent of public deliberation that occurred in setting up the Oregon Health Plan provide a model that was successful in some regards and that can be improved and refined. For example, if the survey data indicate poor ratings for access to treatments, then decisions about allocating resources would benefit from open discussion.

Making Adjustments

Using Assessments as Feedback Information

All the assessment information used in report cards is potentially useful as feedback information for internal adjustments, as well as for "consumer" information. These two functions may be complementary or detracting. For instance, premature release of information to consumers can be more damaging than constructive. We therefore recommend that report cards entail a two-step process. They should first be made available to provider groups for feedback and internal adjustment. A period of weeks or months should be allowed for physicians and institutions to articulate their response and identify steps for change. Thereafter, the report should be made public for other groups, keeping the original assessments unaltered but this time including the assessed party's responses and his or her anticipated remedial steps.

Once assessments and discussions among appropriate groups have identified an area needing change, further focused assessment can provide guidance for specific adjustment. For example, patient dissatisfaction with medical intervention decisions can be followed with inquiry into a range of specific decisions patients would make in given scenarios [66]. Any discrepancies found between reported desires and actual decisions noted from record review can be brought to the attention of physicians, who may then need to examine informed consent procedures to ensure sensitivity to discrepancies. The data could also be used to formulate default guidelines for patients who cannot voice their wishes or to assist in matching patient preferences with actual decisions [72].

Continuing Medical Education

The increased use of assessment information to direct aspects of continuing medical education would allow this form of learning to be more integrated into the continuous adjustment process. Providers and institutions already consider continuing education to be an obligation of routine professional practice [73-75]. Periodic ethics case reviews in regular institutional teaching settings, such as morbidity rounds and grand rounds, can help keep standards of ethical conduct fresh in providers' daily judgments. Mandatory periodic recertification is recommended because it can motivate providers to keep perpetually abreast of the field [72-76].

Peer Mechanisms

Peer mechanisms provide a traditional avenue for professional accountability and, because of their known efficacy and ability to preserve a professional core in medicine, should be continued and augmented. Mechanisms for identifying outstanding physicians, whether by requiring evidence of excellent practice for fellowship in prestigious medical associations or by awarding recognitions for high ethical standards, could be stronger [77, 78]. Specific teaching and explicit professional codes could promote the spirit of collegial critique and common pursuit of high standards. Peer groups, peer consultations, and counseling mechanisms offer corrective or improvement methods that do minimal or no damage to the provider or institution. For example, ombudspersons can help clarify issues for erring physicians and can provide summary information to guide institutional improvements.

Enforcement

Accountability is mostly about setting and assessing standards and requiring justification and responsibility rather than about enforcing regulation. Our stratified model of accountability uses many stabilizing forces, such as professional pride and shame, administrative regulation, political processes, and economic forces such as subscribers or providers exiting an unsatisfactory system. However, although these forces can be robust in maintaining norms, they are weak in enforcing specific issues for specific individual persons or institutions. To be effective, accountability mechanisms must be backed up by the potential for external regulatory enforcement.

Audit

Record review, although a limited method for assessment, has merits as a mechanism to enhance compliance [48-54, 56]. Routine checks are widely used in other fields; activities of the Internal Revenue Service, for example, are among the most widely known. Such checks can be infrequent and still have an effect because individual physicians know that routine investigation is possible. We recommend medical record audits, either random or to follow up on poor assessments from other methods, with in-depth follow-through to include interview of patients, family or surrogates, and physicians. Audit findings should have the potential to lead to disciplinary action.

Decertification and Loss of Privileges

Privileges are withheld at many levels. Entry into and graduation from training programs, provision of certification and licensure, employment, and membership in professional organizations can all be withheld. These "road checks," which should all include assessments of ethical conduct, have been mentioned elsewhere. Similarly, institutional quality control mechanisms already exist through the Joint Commission on Accreditation of Healthcare Organizations and the National Committee for Quality Assurance. Minimum standards should be set by a combination of professional opinion, patient opinion, and public deliberation. Clear failure to meet ethical standards should result in revocation of accreditation.

Litigation

Litigation for failures in accountability is potentially available to all parties. We do not mean to endorse litigation as an assessment method or as a fair compensation method. Both have been suggested; indeed, a national practitioner data bank now exists that includes litigation data [79-81]. We concede that evidence suggests an imperfect correlation of legal suit with the communication skills of a physician, but we note that this is distinct from ethical conduct and certainly from technical competence [82-88]. Similarly, a presumption is that malpractice law provides legitimate compensation to injured parties. However, we note that litigation is differentially available to the wealthy and also usually damages the complainant. We therefore restrict our endorsement of litigation in this context to its use as an enforcement mechanism. For example, litigation could be used to enforce removal of accreditation or licensure for breach of accountability standards.

Conclusion: Next Steps for Accountability in Medicine

To date, the initiative for a reformulated medical accountability has come primarily from administrative groups within the health care industry—chief executive officers, chief medical officers, quality improvement department chiefs, governing boards, and so forth. Implementation of accountability is difficult at best, and without professional standards it is also unguided by physicians' understanding of their own activities. We therefore urge professional groups in medicine to provide standard setting recommendations in all domains of medical practice. The set of recommendations provided here for medical ethics is only one among many sets of needed recommendations for accountability. Much work remains. Once offered, administrative bodies can adopt the professionally generated and endorsed content areas and then proceed with surveys, discussion forums, and feedback and adjustment mechanisms.

We have focused only on accountability of physicians and institutions. To maintain balance, it will be essential to further consider the accountability of other groups in the medical system, such as patients, insurance-providing employers, insurance agencies, the government, and other elected officials. Each group must define its own content areas; for example, insurance agencies should include in their list of content areas confidentiality, nondiscrimination, and access. Professionals can help engender growth in accountability for these groups by showing well-developed accountability in their own locus; by initiating dialogue with other groups, using conferences, commissions, and the like; by challenging them with physician-defined content areas and assessments; and ultimately by demanding accountability in the political sphere.

Appendix

The following are members of the Working Group on Accountability: Harvard Medical School, Boston, Massachusetts: Linda Emanuel, MD, PhD (Chair), Ruth Fischbach, PhD, Edward Hundert, MD, Thomas S. Inui, SciM, Lynn Peterson, MD, David Wilkins, JD; Massachusetts General Hospital, Boston, Massachusetts: David Blumenthal, MD, MPP, John Stoeckle, MD; Harvard School of Public Health, Boston, Massachusetts: Troyen Brennen, MD, JD, MPH; Dana-Farber Cancer Institute, Boston, Massachusetts: Ezekiel Emanuel, MD, PhD; University of Massachusetts, Boston, Massachusetts: Floyd J. Fowler Jr., PhD; Harvard Community Health Plan, Boston, Massachusetts: James Sabin, MD; Children's Hospital, Boston, Massachusetts: Robert Truog, MD; University of Minnesota Law School, Minneapolis: Susan M. Wolf, JD.

The following are consultants to the Working Group on Accountability: Robert Arnold, MD (University of Pittsburgh, Pennsylvania); Daniel Brock, PhD (Brown University, Providence, Rhode Island); Howard Brody, MD, PhD (Michigan State University, East Lansing); Christine Cassel, MD (Mount Sinai Medical Center, New York, New York); Norman Fost, MD (University of Wisconsin, Madison); Steven Miles, MD (University of Minnesota, Minneapolis); Mark Siegler, MD (University of Chicago Medical School, Illinois); Susan Tolle, MD (Oregon Health Sciences University, Portland); Alan Weisbard, JD (University of Wisconsin Law School, Madison).