Clinical problem solving: the beginning of the process.

The diagnostic thinking process is often described as one of hypothesis generation and testing. Yet descriptions of the clinical problem solver's thinking prior to generation of the first diagnostic hypothesis are lacking. This paper reports a study of this phase in students, house officers and registrars. Theories of pattern recognition are seriously questioned. It is shown that these subjects make immediate active interpretative or evaluative responses to initial items of clinical information elicited. These responses cannot be described as 'diagnostic hypotheses', but are important working interpretations. Implications for teaching are discussed.