'Noisy patients'--can signal detection theory help?

Signal detection theory tests an observer's ability to discriminate between signal and noise. Deciding whether or not a patient's symptoms warrant further investigation or treatment is an example of this task in the clinical setting. Noise can exist within the observer--for example, in the brain of a tired or inexperienced doctor--or can arise from an external source such as the patient. Patients can produce external noise by giving numerous unrelated presenting complaints, providing overly detailed accounts of their symptoms, or simply talking too quickly. The more noise that is present, the harder the signal (such as a new disease or a notable change in an old condition) is to detect. Patients in the neurology clinic seem to be 'noisier' than average, perhaps owing to the long duration of their condition in many cases and the relatively high proportion of patients with medically unexplained symptoms. The ability to interpret such 'noisy' histories often underpins the neurological diagnosis. This Review aims to promote the relevance of signal detection theory to the overworked neurologist on the ward or in the clinic and explores strategies to reduce the noise generated both within the brain of the doctor and by patients.