(including, in this instance, the parents of patients); to assist in the study of pathophysiology and etiology; and to guide treatment.
Providing a diagnostic label for a condition gives clinicians a common language that is useful in discussion. Diagnoses such as measles or fetal alcohol syndrome provide clinicians with a term that invokes a concept. On hearing this term, they can immediately call on an array of related facts and concepts that assist them in conveying information in a consistent manner. For the term to be useful to clinicians the diagnosis should be used in exactly the same way by everyone for several reasons: epidemiologic data can then be accurately collected, and patient records transferred from one site to another will contain information that can be interpreted accurately. Diagnoses are also used in order to conduct clinical research, with the long-term goal of understanding the mechanisms that cause a particular illness. For this purpose, it is equally important that diagnoses be consistent and accurate. Finally, diagnoses are used to assist clinicians in selecting appropriate treatments. Quite often, the process of differential diagnosis provides a road map that directs the clinician in the choice of treatment programs that otherwise may be quite different. In the differential diagnosis of acute abdominal pain, a diagnosis of appendicitis will send the patient to an operating room, whereas cystitis will lead to a prescription for medication.
Because of the obvious importance of diagnosis in clinical medicine, during recent years nosologists have stressed the necessity of making the diagnostic process more objective and explicit. This improvement in the diagnostic process has often been implemented through the development of "diagnostic criteria." One of the earliest efforts in this direction was achieved by the American Heart Association, which developed criteria for defining the stages of cardiac failure. The most comprehensive effort produced to date was initiated by the Task Force on Nosology of the American Psychiatric Association, which completed a comprehensive revision in the third edition of its Diagnostic and Statistical Manual (known as DSM-III), published in 1980. In this manual, for the first time a comprehensive list of diagnoses for mental illness was defined through the use of explicit diagnostic criteria. This process led to a careful conceptualization of the best way to develop diagnostic criteria for specific illnesses.
Two requirements are typically used to evaluate diagnostic criteria. First, the criteria must be reliable. Second, they must be as valid as possible.
The concept of reliability refers to agreement among clinicians and consistency over time. In both cases, the essential feature is that the diagnosis be consistent and stable, so that good agreement occurs. Interrater reliability refers