Supplementary Material for: The feasibility of using ultra-widefield retinal imaging to identify ocular pathologies amongst those with systemic medical disease attending tertiary healthcare facility at a university hospital

Aims To evaluate the feasibility of ultra-widefield (UWF) imaging to identify ocular pathologies amongst in- and out-patients in a tertiary university hospital. Methods Prospective double-blinded multicenter clinical study. In total, 634 patients from a University Hospital with pulmonary, cardiovascular and endocrine diseases were examined by two teams by conventional slit-lamp biomicroscopy (CBM). UWF images with Optos Tx200 were recorded and graded independently by two retina specialists, and graders from two Reading Centers for presence of pre-defined pathologies. Interrater reliability was calculated using Fleiss statistical software. BL classified all UWF images with retinal hemorrhages by severity and interrater agreement. Results Complete data were available for 502 patients. Reading Center Moorfields Eye Hospital, London, UK (RM) reported the highest number of cases with retinal pathologies (378), Reading Center GRADE Bonn, Germany (RB) did so for cases with optic disc cupping (466), R1 with optic disc pallor (151) while R2 reported the lowest number optic disc pathologies (39). Interrater reliability was highest for retinal hemorrhages (0.59). To understand low interrater reliability, sub-analysis of retinal hemorrhages indicated high interrater agreement for obvious pathologies. Conclusions UWF imaging is convenient and effective to identify retinal pathologies in patients attending hospital clinics for a range of systemic diseases. Imaging the eye allows for remote assessment and reassessment of the retinal diseases, however the time taken to train graders and for image analysis and reporting should not be underestimated.