Medical Research, Medical Journals and the Public Interest
Relman, Arnold S., Journal of the Society of Research Administrators
MEDICAL RESEARCH, MEDICAL JOURNALS AND THE PUBLIC INTEREST
The health of medical research is essential to the health of the American people and of the world. Without a broadly based vigorous program of research in basic science and applied clinical science, there is no progress. Everything we know now, everything that we can do, everything that medicine has to offer in the way of improving health is based on research. Everything that we hope to achieve in the future, dealing with all the terrible health problems all over the world, depends on progress in research. Those who imagine that we can get along without supporting medical research are deluding themselves.
Medical research is an enterprise that is carried out in many kinds of institutions by many kinds of people, with various kinds of support. Important research is done in academic institutions, government institutions, and independent research institutions. Regardless of institution, the process is the same. It begins with investigators getting an idea. And, if they have the resources, they investigate that idea; they begin to carry out experiments. The experiments may change the hypothesis on which the research was based. The initial experiments may indicate that the idea was wrong, and they have to go off in another direction. Or they may get some fruitful results, but then, because of what happens later on, the direction of the research may change. But sooner or later, with any luck, they have some data that they think mean something, and they try to draw some conclusions. They talk to their colleagues; they discuss their work at seminars and private meetings that are held among the scientists working in the field. As a result of those discussions, their ideas may change, and in comparing notes with colleagues they may learn something that causes them to change what they are doing--perhaps to drop one line of investigation to try another. They may be asked to present their work at a local or international scientific meeting. At the time the work is presented, it may be finished and ready to be written up and reported in the medical literature or it may still be in progress.
Scientific meetings, then, are simply occasions for working scientists to talk to one another. More than half of the papers reported at medical scientific meetings never get published, i.e., the work never comes to fruition. Why? Because it turns out that the original interpretations were wrong, the methods weren't appropriate or, as the result of further work, the conclusions presented at the meeting were not quite correct. This fact should not be a cause for concern, since science is based on corrections of error and testing of hypotheses. Those misinformed bureaucrats who think that they have to police science to eliminate error, or who confuse error with fraud, don't understand that science advances through the postulation of hypotheses which may turn out to be right or wrong; collection of data, which may be adequate or inadequate; correction, discussion, and rigorous criticism of the data and the postulates; postulation of new ideas; collection of additional data; and so on in a continuing process. The point is that the best available information is often simply a way-station to the ultimate answer. One never knows in advance what the answer is because, often, as methods change and ideas change, new information becomes available and new insights are put forward.
At some juncture, however, a working scientist decides that there are enough data to fit the story together. The scientist then writes it up in the form of a research report to submit, usually to a peer-reviewed journal. Then the journal editors send that paper out to experts in the field for criticism and review. At The New England Journal of Medicine our custom is to send manuscripts to at least two experts in the field--recognized authorities who are working on that subject themselves. If the manuscript contains numerical or statistical analysis, we always add a third reviewer who is a biostatistician or epidemiologist. …