The influence of image interpretation training on the accuracy of abnormality detection and written comments on musculoskeletal radiographs by South African radiographers

Introduction The study was conducted in two government hospitals in Johannesburg, South Africa, investigating the extent to which image interpretation and relevant terminology training would improve the accuracy and descriptive comments provided on musculoskeletal images by South African radiographers. Methods Nine radiographers interpreted an image bank comprised of 100 skeletal radiographs (50% abnormal) both before and on completion of a tailored education program in image interpretation. Radiographer comments were compared with the reference standard diagnosis (single experienced radiologist) and deemed to be correct, partially correct, or incorrect. The radiographers were assessed for sensitivity, specificity, and accuracy on the image bank pre- and postintervention. After testing for normality of the data, a Wilcoxon signed rank test was used for nonparametric paired data. Results Radiographer accuracy (from 71.04% to 78%), sensitivity (from 83.73% to 87.28%), and specificity (from 59.62% to 70.34%) all improved after the education program. A statistically significant improvement was noted in the accuracy (Wilcoxon value, z = -2.66, P =.008). Incorrect radiographer comments also decreased (from 24.1% to 17.78%, Wilcoxon value, z = -1.96, P=.05). Radiographer vocabulary used when describing abnormalities was more in line with the reference standard diagnosis after training. Conclusions This study has shown that training in pattern recognition and construction of a comment could enable diagnostic radiographers to improve their accuracy and the ability to provide a descriptive comment on an image. Future work should include assessing accuracy and commenting in the clinical environment and whether the improvement in commenting is maintained over time.