Abstract
Purpose: :
To measure accuracy of retinopathy of prematurity (ROP) diagnosis by pediatric ophthalmology fellows.
Methods: :
An atlas of 804 retinal images was captured from 248 eyes of 67 premature infants with a wide-angle camera (RetCam-II; Clarity Medical Systems, Pleasanton, CA). Images were uploaded to a study website, from which an expert pediatric retinal specialist and four pediatric ophthalmology fellows independently provided a diagnosis (no ROP, mild ROP, type-2 ROP, or treatment-requiring ROP) for each eye. Sensitivity and specificity of each fellow were calculated compared to a reference standard of diagnosis by the expert pediatric retinal specialist.
Results: :
For detection of mild ROP or worse, mean (range) sensitivity was 0.926 (0.838-0.962) and specificity was 0.874 (0.832-0.902). For detection of type-2 or worse ROP by fellows, mean (range) sensitivity was 0.637 (0.465-0.907) and specificity was 0.905 (0.732-0.985). For detection of treatment-requiring ROP, mean (range) sensitivity was 0.750 (0.267-1.00) and specificity was 0.911 (0.764-1.00).
Conclusions: :
Pediatric ophthalmology fellows in this study generally demonstrated high diagnostic specificity in image-based ROP diagnosis. However, diagnostic sensitivity was lower, particularly for clinically-significant levels of disease. This raises concern for under-diagnosis of disease and suboptimal management of a potentially blinding disease. This has implications about the quality of ROP screening by fellows and other less experienced ophthalmologists, particularly when performed without supervision, and suggests a need for formalized training in ROP assessment.
Keywords: retinopathy of prematurity • learning • clinical (human) or epidemiologic studies: health care delivery/economics/manpower