from each other about more and less successful ways of handling particular situations.

• Accountability. Does the conflict of interest policy indicate who is responsible for monitoring, enforcing, and revising it? Leaders of accountable institutions explain institutional policies and monitor and accept responsibility for the consequences, both beneficial and harmful.

• Fairness. Does the policy apply equally to all relevant groups within an institution and in different institutions? In an academic medical center, the relevant groups would include faculty, medical staff, students, residents, fellows, members of institutional committees (e.g., institutional review boards, formulary committees, panels developing practice guidelines, and device purchasing committees), and senior institutional officials.

Concerns about conflicts of interest in medicine have a long history, and responses to these conflicts have evolved as relationships with industry have grown more frequent and more complex and as different responses to such relationships have been tried and found in need of modification. Government regulations and voluntary codes of conduct often follow the discovery of instances of questionable or inappropriate relationships and conduct. Government scrutiny of financial relationships and conflicts of interest may also stimulate private, voluntary efforts by academic and other institutions to deal with problems and avoid regulation.

The conflict of interest policies of academic medical centers, professional societies, medical journals, and other institutions vary on many dimensions. It is not clear that all medical institutions have conflict of interest policies. Those that do have such policies vary in what they ask physicians and researchers to disclose about their financial relationships with industry. Such variations may create additional administrative burdens for physicians and researchers who act in multiple roles and make multiple disclosures of their financial relationships with industry to different institutions for various purposes related to medical research, education, and clinical care and clinical practice guideline development.

Institutions also vary in what relationships they prohibit because they view them as creating unacceptable risks of undue influence on primary interests, and they also differ in how they manage conflicts of interest that are not prohibited. The National Institutes of Health (NIH) has identified variations and deficiencies in how research institutions implement the 1995 U.S. Public Health Service (PHS) regulations on conflict of interest, and it has advised institutions on steps that they can take to strengthen their poli-