Diagnostic accuracy would be improved by developing more categories of "non-disease".

Studies on the ways in which doctors think have shown the importance of intuitive processes in the formulation of a diagnosis. They have also suggested that intuitively generated diagnoses are used to aid the memory of relevant clinical facts during a clinical consultation. Only some symptoms and signs can be ordered into or subsumed under a diagnostic label. Others, which have not been thought of as constituting a disease, will tend to be lost from memory. If a diagnosis were to be thought of, not as the label of a disease but as a mental construct by means of which data are ordered and remembered, that is, as an invention whose utility is to be judged by the degree to which clinical thinking is facilitated, it would become reasonable to develop diagnoses to cover larger areas of clinical phenomenology. Some examples of the usefulness of such diagnoses of "non-disease" are given ant it is suggested that clinical thinking might be improved by the wider use of the concept.