Aims: To identify whether after performing retinal photography, direct ophthalmoscopy can improve the yield of screening for the detection of sight-threatening diabetic eye disease (STDED).
Methods: Patients (n = 408) who had previously received both dilated direct ophthalmoscopy by a diabetologist and retinal photography graded by a diabetologist within 3 months of each other were included. The results of the other screening modality were not available to the grader/screener. The first 308 patients were consecutive attendees at the clinic who fulfilled the study criteria and 100 were selected because they were identified as having potential STDED by either one of these modalities. An ophthalmologist using slit lamp biomicroscopy then examined patients identified with potential STDED.
Results: In 357 (88%) patients there was agreement between the two modalities about whether referral to an ophthalmologist was required (kappa 0.62). Retinal photography identified 38 patients for referral to ophthalmology which ophthalmoscopy missed. Of these, the ophthalmologist agreed that STDED was present in 32 (84%) and four patients required early laser. Ophthalmoscopy identified 13 patients for referral who were not identified by photography. Of these, the ophthalmologist agreed with the diabetologist that STDED was present in seven (54%) and one patient required early laser.
Conclusion: Ophthalmoscopy may identify the occasional patient with diabetes who has STDED which is missed by retinal photography. For a systematic retinal screening programme, adding ophthalmoscopy to retinal photography will increase false-positive referrals and is likely to detect only a few extra patients requiring laser.