To the Editor: As chairperson of APA’s Committee on Confidentiality, I am responding to the three articles regarding researchers’ use of psychiatric records (1–3). There is no question about the value of epidemiological research. However, as more and more uses are sought for the expanding database of patient information, it is important that organized psychiatry carefully consider the clinical, ethical, and research implications of these developments. Notions that we must balance research needs and privacy (1) and combat stigmatization by not allowing too much privacy protection (2) pit privacy against other laudable aims. Workable policies for database research must incorporate, rather than compete with, patient privacy. This principle was bolstered by the Supreme Court’s rejection of the balancing test for psychotherapy privilege in Jaffe v. Redmond (1996) " because it would eviscerate the effectiveness of the privilege" (4).

Furthermore, the reliability and accuracy of research using health plan or insurance claim data is questionable owing to the current practice of physicians manipulating reimbursement rules (i.e., exaggerating patients’ conditions, changing billing diagnoses, and/or falsely reporting patients’ symptoms) in order to secure needed coverage for patients. This, too, must be addressed. It is easy to imagine that physicians’ and patients’ privacy concerns could increase this unfortunate trend.

I agree with the recommendation of Paul S. Appelbaum, M.D. (3), that patients’ consent should be obtained prospectively when information is collected for future research purposes. Gregory E. Simon, M.D., and colleagues (2) appeared to support this approach when they suggested that informed consent should be obtained for the use of clinical data whenever possible.

I also agree that the solution is less clear-cut when we are dealing with existing records or databases that were not originally collected or intended for research purposes. But I disagree with a policy that allows researchers to use existing identifiable patient records after receipt of only institutional review board approval. Dr. Appelbaum considers a possible policy of obtaining patients’ prospective blanket consent for research, and he is aptly troubled by the absence of true informed consent. But leaving patients completely out of the loop by using institutional research board approval as a fallback mechanism fails to solve this dilemma, because it deprives patients of the opportunity to choose at all.

Such a policy would allow our patients’ clinical records, created for their care, to be viewed by researchers without regard to patients’ right to privacy in treatment. This is a dangerous precedent, because it would destroy people’s trust in psychiatry and psychiatrists. We now have strong evidence that patients’ privacy concerns result in thwarted communication and tainted data (5). This adversely affects both care and research.

Ethically, that kind of policy ignores some key principles in the Nuremberg Code (1947) and the Declaration of Helsinki (1964), in which respect for individual autonomy was to supercede "scientific and societal goals" (6). The National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research was also concerned about the implications of using medical records for research and defined that practice as research with human subjects (7). In its 1978 Belmont Report (8) the commission recommended that the rights of subjects were to be respected, specifically by obtaining their informed consent to participate.

But the commission members, too, recognized the difficulties of obtaining informed consent from each patient for each research use of his or her records, often years after the patient had left a health care institution or provider. Thus the commission recommended patients be given the opportunity at admission to opt in or out of the research use of their medical records (9).

I suggest an alternative policy for the protection of privacy in research, based on the commission’s 1978 reports (8, 10) and developed by the National Coalition for Patient Rights (9) For planned prospective uses of patient records, informed consent should be sought. However, for retrospective research using patient records, the model suggests that prospective consent should be obtained when patients enter a health plan or health care setting. To ensure an added layer of protection, this consent is delegated to a board specifically designed to review these matters: a medical records review board. Such a board would relieve already overworked institutional review boards of an additional burden, would not suffer the institutional conflicts of interest inherent in institutional review boards, and would contain more community representation than institutional review boards. In their article Dr. Simon et al. alluded to this kind of policy when they noted that "health insurers have numerous opportunities to communicate with members regarding research uses of health care data" (2, p. 1735). They suggested that abbreviated or simple consent procedures be incorporated into routine practice and that they might be "preferable to complete waiver of consent" (p. 1735).

Clearly, this is an imperfect policy, since it does not address the problem of records that currently exist and are being sought for research. Had the U.S. Department of Health, Education, and Welfare followed the original recommendations of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, we would not be faced with this dilemma today. But if the new policy were established as I suggest, we could deal with existing records by attempting to contact patients to obtain either study-specific or delegated general consent for research. If that is not possible, the matter could be delegated to a medical records review board capable of making binding decisions. This would be preferable to giving responsibility to an institutional review board, because the medical records review board would contain more community representation and be less susceptible to institutional pressure to approve research proposals.

Patient records created after the date of the policy implementation would require "prospective delegated consent," with the added provision that patients be given the chance to revise their decisions when they return for care or to change health plans or be practicably informed about a proposed study using their records. With this policy in place, fewer and fewer patient records would be used without individuals’ knowledge or consent.

This approach, with its additional steps, would allow valuable research to proceed while respecting and protecting patient privacy and dignity. This would foster trust and, because it would encourage openness and honesty with clinicians, would likely advance the kind of crucial research to which we are all committed.