The use of risk factors in medical diagnosis: opportunities and cautions.

We discuss in this paper the extent to which disease risk factors may assist in the diagnostic process. We caution that disease risk factors need not be very sensitive or specific. Risk factor specificity and sensitivity may be further reduced if, in the former case, the risk factor is related to other illnesses having the same clinical presentation as the disease of interest, or if, in the latter case, the risk factor disappears with the onset of illness. We illustrate these points in a discussion of the utility of smoking as a diagnostic test for malignancy in two clinical situations, the patient with asymptomatic microscopic hematuria and the patient with a solitary pulmonary nodule. Risk factors hold great promise as aids to medical diagnosis, as this information is readily available to clinicians at little or no cost. Clinicians, however, should exercise caution when using risk factors of unproven diagnostic utility in medical diagnosis, as their presence may have little or no effect on disease probability.