Standard method of diagnosis versus use of a computer database in the evaluation of skeletal dysplasias

The objective of this study was to compare reference textbooks and the computer database, OSSUM, for accuracy and ease of use in the diagnosis of skeletal dysplasias. Materials and methods. Twenty cases of clinically and and radiologically established skeletal dysplasias were evaluated as unknowns by four pediatric radiologists. Readers 1 and 2 evaluated group A (10 cases) using reference texts and group B (10 cases) using OSSUM. Readers 3 and 4 evaluated group B using reference texts. The radiologists independently listed their roentgenographic findings, the top three diagnoses, confidence level, difficulty level, and time spent on each case.

The correct diagnosis was made in 68% of both the reference text cases and the OSSUM cases. Difficulty level was significantly higher (3.5 vs 2.9, P = 0.0013) and confidence significantly lower (3.3 vs. 2.3, P = 0.0001) when using OSSUM. Average time spent on cases was 25 min with references and 30 min with OSSUM (P > 0.05). However, there was a decrease in both the time (38 min vs 23 min, P = 0.05) and the difficulty (3.9 vs 3.1, P = 0.001) between the first five and the last five cases. The composite of four readers correctly identified 90% of the skeletal dysplasias when the results of both methods were combined.

In the ability to reach a correct diagnosis, no difference was detected between the OSSUM and reference texts methods. The increased time necessary, greater difficulty and decreased confidence levels with OSSUM are expected to improve with increasing program familiarity. Use of both textbooks and the database was complementary.