GOVERNMENT HEALTH AGENCIES now can, and do, keep detailed health records of millions of Americans. By means of computers and microfilm, information including cause of death, occupational exposures, diet, drug use, reproductive history and so forth, can be stored on a scale unthinkable until now. There are huge listings for cancer and occupational disease, and there is now a U.S. National Death Index. This mass of information can be a researcher's gold mine. But the existence of the files also poses ethical and legal questions that have not been resolved.

The particular advantage of such data is that it can pinpoint answers before anyone knows how to ask the right questions. Asbestos, for instance, was found to be the cause of disabling lung disease through the study of this kind of information, rather than through the study of the condition itself. As in this case, or in the case of most cancers, if the biological causes of a disease are unknown, or if there is a long latent period between exposure to the disease-causing agent and appearance of the disease, such data banks can be the only way to uncover the link between a disease and its cause.

These medical records are thus a powerful tool whose potential has only just begun to be tapped. But their existence may also impose ethical responsibilities beyond -- and often in conflict with -- those associated with medical research. If a particular chemical or type of behavior is found to be hazardous, must those whose names are in a file be informed? What if the connection is only tentative, or if there is nothing the individual can do about it except to worry? Is it doing anyone a favor to provide such information, especially when stress itself is known to be a powerful contributor to sickness? Is it ethical to provide notification without also providing medical follow-up? And if that is required, what if the costsof providing it are so great that it becomes impossible to collect the information in the first place? Then clearly, everyone loses.

Deciding who should have access to the data raises a different set of questions. Government agencies collecting this kind of information have an unusual exemption from the Internal Revenue Service law that otherwise keeps taxpayers' records strictly private, but they do not have a corresponding exemption from the Freedom of Information Act.FOIA requests can easily invade personal privacy that medical researchers normally take great pains to protect. Premature release of data -- by those with business or other interests to pursue -- can cause public confusion, fear and economic losses.

In the wake of some notorious cases of government's irresponsibly declining to notify people of ailments inflicted onthem -- and often inflicted by government action -- there is a tendency, which we share, to view such secret records keeping with suspicion. There is also something naturally repugnant in the government's knowing that an individual was exposed to a likely carcinogen and not telling him or doing anything else about it. But no one has yet figured out the proper guidelines for disclosure and notification where large-scale medical records are concerned. Maybe a limited-access, research-only policy is the wisest course. Certainly this is one rare case where a government task force could do some good.