Context: The language that patients use to communicate with doctors is quite different from the language of diagnosis. Patients may describe tiredness and swelling; doctors, fatigue and oedema. This paper addresses the process by which novices, who have learned standard medical terms for symptoms, use lay descriptions of symptoms to reach a diagnosis. Data in this paper indicate that the familiarity of the language used to describe symptoms influences diagnosis in novices and diagnosis does not, therefore, involve a simple translation into standard terms that are the basis of diagnostic decision.
Methods: A total of 24 undergraduate students were trained to diagnose 4 pseudo-psychiatric disorders presented in written vignettes. Participants were tested on cases that contained 2 equally probable diagnoses, in 1 of which the symptoms were expressed using previously seen descriptions. A deviation from 50:50 in reported diagnostic probabilities was expected if the familiar symptom descriptions biased diagnostic decisions. Twelve participants were tested immediately after training and 12 after a 24-hour delay.
Results: Participants assigned greater diagnostic probability to the diagnosis supported by the familiar feature descriptions (F[1.242] = 19.35, P < 0.001, effect size = 0.40) on both immediate (52% versus 41%) and delayed (51% versus 38%) testing.
Discussion: The findings indicate that diagnosis is not simply based on a process of translating patient descriptions of symptoms to standard medical labels for those symptoms, which are then used to make a diagnosis. Familiarity of symptom description has an effect on diagnosis and therefore has implications for medical education, and for electronic decision support systems.