Training evaluation of a course in diabetic retinopathy screening

Abstract

Abstract

The success and effectiveness of diabetic screening programmes are dependent on the availability of appropriately trained image graders. This study was designed to evaluate graders enrolled on a locally devised, formal training course by means of a performance-based measure. The course consisted of four days of classroom-based tuition followed by three months of practice-based learning in the workplace. The aim was to establish whether trainees showed an improvement in their ability to grade images, and secondly whether test sets of images are useful in measuring training outcome. Thirteen trainees were required to grade a test set of 24 single images both before and after training. A significant improvement in sensitivity (from 35% before training to 45% after training) was observed as a result of training but at a cost of a decline in specificity. Trainees’ confidence ratings measured on a five-point scale increased from an average of 2.4 to 4.1 (p<0.01). We concluded that the course needs to focus more on trainees’ ability to discriminate between normal and abnormal images as well as improving grading accuracy in line with increased grading confidence. Test-based course evaluation can be seen to be a valuable instrument in establishing a quality standard for stated learning outcomes. In this research it has clearly indicated weaknesses of the training programme in its current form.