Can the diagnostic accuracy of newborn eye screening for congenital cataract be improved with digital imaging? The Digital Imaging versus Ophthalmoscopy (DIvO) study

Abstract

Background Congenital cataract is the commonest cause of preventable child blindness in the world. It affects 1: 3000 babies, 60% of whom will have bilateral disease. Although screening of the newborn red-reflex with an ophthalmoscope is recommended, there are questions over this technique's accuracy, particularly in non-specialist hands. Several studies in enriched paediatric eye clinic populations have demonstrated superior accuracy when a digital image of the the eye's reflection to infrared light is evaluated. Aim This study uses a prototype infrared digital camera to test the hypothesis that the sensitivity and specificity of newborn screening using evaluation of the "infrared-reflex" image is superior to the standard red-reflex examination using ophthalmoscopy. Methods 140,000 newborn babies will be recruited into the study from at least 13 maternity units in England over an 18 month period. Babies will have both the standard red-reflex assessment and evaluation of the infrared-reflex using a prototype device. Since specialist gold-standard evaluation of every participant is impractical, bespoke data linkage requests to NHS England will be made. Data from the red-reflex evaluation and the presence of codes relating to cataract diagnosis and/or treatment in Hospital Episode Statistics (HES) will be retrieved for each participant a minimum of 6 months after their birth. This data will be used to calculate relative and absolute sensitivity and specificity for each screening test and comparison of accuracy using the McNemar test. Secondary outcome measures will include comparison of accuracy in different ethnic groups and screener usability scores. Anticipated impact Confirmation of the hypothesis will support development of a commercial screening device and the possible revision of newborn screening recommendations. Trial registration The study is registered with clinicaltrials.gov: NCT05282147.