Graphic representation of clinical symptoms: a tool for improving detection of subtle fractures on foot radiographs.

Abstract

The purpose of this study was to assess changes in accuracy, degree of confidence, and evaluation time in radiography of subtle foot fractures when the text history is supplemented by a graphic indicating the site of pain. Radiographs from 226 foot examinations (three views), including 126 examinations showing one subtle fracture (< 1-mm displacement) and 100 examinations with normal findings were selected. In the first interpretation session, only a text history was given for 112 examinations, and both text and a graphic indicating the site of pain for 114 examinations. Six months later, a graphic and text history were provided for the 112 cases interpreted without a graphic in the first session, and only text was provided for the other 114 cases. Seven radiologists evaluated the study sets. Sensitivity, specificity, degree of confidence (1-10 scale), and mean interpretation time in seconds were calculated. Use of a graphic increased overall sensitivity for any subtle fracture from 67% to 73% (p < 0.001), increased degree of confidence from 8.1 without a graphic to 8.4 with a graphic (p < 0.0001), and decreased the time for interpretation by 6%, from 53 seconds without a graphic to 50 seconds with a graphic (p = 0.006). Specificity changed from 93% without a graphic to 94% with a graphic (p = 0.33). Fractures of the third metatarsal were missed most frequently (74%); this percentage improved to 61% with use of a graphic. A graphic complements the text history by improving sensitivity, degree of confidence, and time for interpretation.