To the Editor: We worry that Dr. Goldman’s letter will perpetuate unfounded fears regarding research uses of medical records. To illustrate, consider a study examining the long-term effects of psychotherapy on mortality due to alcohol abuse. Such a study would require linkage of large databases containing information on receipt of psychotherapy (insurance claims records) and causes of death (state vital statistics records). While collection of identifying information would be necessary for this electronic linkage, there would be no conceivable scientific need for examination of individual patient records. Any attempt by research staff to view individual records would be clearly inappropriate—and grounds for disciplinary action. We use this example to illustrate an essential point: quantitative research considers the average experience of large (and anonymous) groups rather than the particular experience of identifiable individuals. Research using computerized records should not allow "patients’ clinical records, created for their care, to be viewed by researchers without regard to patients’ right to privacy"—at least not in the usual sense of the word "viewed."

In her reference to legal decisions regarding psychotherapist-patient privilege, Dr. Goldman blurs a distinction that we had hoped to emphasize. Psychotherapy records are protected from subpoena for good reason. Adversarial legal processes are always concerned with the particular experiences of identifiable individuals. In the context of litigation or criminal proceedings, invasion of privacy is intentional (and often malicious) rather than accidental. Those requesting records often do so with the clear intent of harming our patients’ or clients’ interests. The methods and aims of research could not be more different. We draw paradoxical reassurance from the knowledge that, to the epidemiologist or health services researcher, each patient really is just a number. We should reemphasize the same distinction between anonymous research and other uses of health care information that focus on individuals (e.g., utilization review).

We share Dr. Goldman’s concern that providers’ attempts to influence coverage or reimbursement may undermine the validity of claims data. Incentives for miscoding may have large or small effects depending on the research questions and methods. Potential biases due to miscoding should be carefully considered by researchers and by institutional review boards evaluating the scientific merit of research proposals. We should add, however, that inconvenience to researchers is one of the less important ill effects of our deficient system for financing mental health care.

We agree with Dr. Goldman that patients should be allowed to opt out of records-based research. When research use is anticipated, consent should be obtained before data are collected. We do not, however, concur with her proposal that decisions regarding research use of existing data be delegated to a medical records review board. The supposed merits of such a board (community representation, independence, and authority to make binding decisions) are all legally required characteristics of institutional review boards. If existing institutional review boards fail to meet this standard, the appropriate solution seems to be reform of institutional review boards rather than creation of a separate review process for records-based research. Records-based research deserves the same level of ethical review as traditional clinical research—not more or less.

Reprints are not available; however, Letters to the Editor can be downloaded at http://ajp.psychiatryonline.org.