EDITORIAL The Place of Epidemiology in the Study of Rheumatoid Disease T HE METHODS of epidemiology are appropriately looked on as at the extreme, opposite end of the methodologic scale from those of biochemistry and biophysics, for epidemiology deals with disease at the population level rather than the individual, cellular, or molecular level. In general, epiclcmiology may be said to have three important functions. The first is to extend the understanding of the severity gradient by studying those persons who have the disease in a form too mild to cause them to seek medical treatment. In the second place, it can provide direct information about the extent of the problem of any particular disease and about variations in the extent of this problem with respect to time and space. In the third place, studies of populations can provide useful information about association between various factors of interest and the presence or absence of a particular disease. With these thoughts in mind, let us take a brief look at the contributions that epidemiology has made in recent years to our understanding of rheumatoid disease. There seem to be three ways in which field studies have contribllted to our understanding of its gradient of severity. The first of these is the renewed interest in morning stiffness as a unifying factor in rheumatoid disease. The importance of this has been emphasized in the reports of the Pittsburgh Arthritis Study. The second contribution to our knowledge of the yraclient comes from the recognition of cases of rheumatoid disease without arthritis. We have now seen rheumatoid nodules without arthritis, or rather preceding the development of arthritis, x-ray evidence of rheumatoid arthritis vreceding any joint complaints, and serologic tests positive in patients with arteritis without arthritis. The third contribution in this area is best expressed hy the attitude of Kellgren' and his colleagues in England that the sheep cell test is an indication of rheumatoid arthritis or of susceptibility thereto. This last concept brings a new tool to the study of the familial aspects of this disease, for it was recently reported by Lawrence1 and by Ziff' that the serologic tests are much more commonly positive among the siblings of patients with rheumatoid arthritis than they are in the general population. This matter is clearly of great potential importance and deserves further seri011s attention. In terms of estimates of incidence and prevalence, we can only say that to date we haveno adequate estimates of the total incidence of the disease though we do have some estimates of the incidence of disease severe enough to cause the individual to seek medical attention. With regard to prevalence, on the other hand, considerable progress has been made in our ability to make comparisons in time and space as a result of the acceptance of the American Rheumatism Association Criteria.2 Doubtless, all of us have some reservations abont the absolute validity of these criteria; however, few would dispute the fact that the existence of these criteria markedly reduces inter-observer variation in the diagnosis of this disease. The first step towards establishing a geographv of rheumatoid disease w a s accomplished with the piblication of 253 254 EDITORIAL a report on the difference in prevalence between Pittsburgh, Pennsylvania and Manchester, Englande3 Up until quite recently, the tools have hardly been available for the thirdmentioned aspect of epidemiologic work, namely, the study of the association between rheumatoid arthritis and assorted host and environmental factors. AS a result, most of the studies that have been done are based on full clinical examination of samples of individuals. From studies of this sort have come some indications that marital status is of importance in regard to this disease, evidence that infections may be associated with onset, and a partial clarification of the relationship between rheumatoid arthritis and osteoarthritis. In general, however, it is felt that this method is much too laborious for widespread application. For studies of this sort there seem to be three possible methods for circumventing the difficulties of full clinical examination. The simplest, and least expensive, though perhaps conceptually the most difficult to accept, is the technic proposed by Rubin and his collaborator^.^ This technic involves the use of a set of questions asked by a non-medical interviewer. The answers to these questions constitute an index of rheumatoid arthritis of sufficient sensitivity and specificity to give useful results when samples of sufficient size are interviewed, and as long as there is reason to believe that the index has the same validity among persons with the characteristic under study as among persons without the characteristic in question. A practical demonstration of the usefulness of this method is found in the report of King on Psychosocial Factors in the Epidemiology of Rheumatoid Arthritis5 The second way to circumvent our difficulty is through the use of serologic surveys of the kind already done by Feldman in Syracuse. Serologic surveys tend, unfortunately, to be *moresubject to non-response errors because there are a reasonable number of people in the general population who are unwilling to give a blood sample. The technic is, however, extremely useful. The third possibility is the development of a screening device of sufficient sensitivitv which can be applied to people in their homes or at their places of work bv nurses, medical students, or even possibly by entirely non-medical persons. Those who screen positive are then referred for detailed examination, serologic tests and x-ray examination in a clinic. This clearly cuts the expense of examining everybody and reduces to a minimum the non-response error because, by and large, those who are found to have some evidence of arthritis on a screening examination are willing to have it further investigated in the clinical situation, and this type of interview is usually well accepted. A technic of this sort is now being tried in the Pittsburgh Arthritis Study and the screen currently in use shows great promise. With the field methodology for epidemiologic study of this disease developing SO rapidly, and analytic methods advancing also,l it now seems highly probable that this investigative discipline will take its place along side of, and share the honors with, the laboratory and the clinic in the years to come. The field is wide open, for there are very few physicians interested in this approach. The basic training required for this kind of work is usually obtained in schools 255 EDITORIAL of public health. At some of these schools substantial fellowships are available to individuals seeking training in the epidemiology of this and other noncommunicable diseases. SIDNEYCOBB,M.D., M.P.H. REFERENCES 1. Kellgren, J. H., Ed.: Transactions of the First International Conference on Population Studies in Rheumatoid Arthritis. In press. 2. Ropes, M. et al.: Proposed diagnostic criteria for rheumatoid arthritis. Ann. Rheumat. Dis. 16:118, 1957. 3. Cobb, S. and Lawrence, J.: Towards a geography of rheumatoid arthritis. Bull. Rheumat. Dis. 7:133, 1957. 4. Rubin, T. et al.: The use of interview data for the detection of associations in field studies. J. Chron. Dis. 4:253, 1956. 5. King, S. H. and Cobb, S.: Psychosocial factors in the epidemiology of rheumatoid arthritis. J. Chron. Dis. In press.