Hypercard coding system for the ACR Index for Radiological Diagnoses

We describe a computer version of the Index for Radiological Diagnoses of the American College of Radiology (ACR). This system combines a graphics interface with a search mechanism while preserving the hierarchical structure of the Index. The graphics interface allows easy selection of an anatomic part, while the search mechanism provides the code numbers associated with an entered term. The computer system and the paper-back version of the ACR Index were compared by having 52 volunteers (21 radiology faculty members, 21 radiology residents, and 10 medical students) each code 30 cases with the book and a matched set of 30 cases with the computer. The average time to code cases was shorter when the computer was used (52.2 vs 64.5 sec; p < .0001). Accuracy was higher when the computer was used (96.4% vs 90.4%; p < .0001). The average confidence in the computer diagnosis was also higher (9.73 vs 9.51, on a scale of 1-10; p = .0016). This system demonstrates the ability of a computer program to outperform an analogous noncomputerized system.