A 'fail-safe' screening programme for diabetic retinopathy.

OBJECTIVE

To improve screening for diabetic retinopathy in a hospital diabetic clinic through the use of the audit process.

DESIGN

Comparison of an existing system of screening for diabetic retinopathy (a specialist optometrist using ophthalmoscopy alone) with a new system in which a specialist optometrist examined retinal Polaroid photographs taken through pharmacologically dilated pupils and combined this with ophthalmoscopy in all cases except when the photographs were perfect and definitely showed no retinopathy. In this new system, the optometrist could discuss cases of uncertainty with a diabetes physician while the patient was still in the clinic with eyes dilated.

SETTING

Inner city hospital diabetes clinic.

SUBJECTS

289 hospital diabetic clinic patients not already attending an ophthalmologist; a consecutive series of 144 such patients for the first audit, 145 for the repeat audit.

MAIN OUTCOME MEASURES

Assessment of each screening system against a gold standard. For the first audit this was agreement by two of four diabetes physicians, who combined examination of the photographs with the findings from dilated ophthalmoscopy, on the classification of the retinae of each patient, guided by standard European criteria. For the second audit, the gold standard was enhanced by discussing the photographs and findings of all patients with an independent ophthalmologist. For patients requiring referral, a second ophthalmologist also commented on the case.

RESULTS

The addition of retinal photography to universal pupil dilatation, and the availability of diabetes physician backup to discuss cases of uncertainty, greatly increased the optometrists' detection rate. Sensitivities for the first (ophthalmoscopy only) and second (ophthalmoscopy plus photography plus diabetologist back-up) audits were, respectively, 71.4% vs 100% for sight-threatening retinopathy, 33% vs 100% for retinopathy requiring six-month review, and 40.3% vs 97.2% for any retinopathy (p = 0.002).

CONCLUSIONS

Optometrists specialising in diabetic retinopathy using Polaroid retinal photography and ophthalmoscopy, both through dilated pupils, backed up by experienced diabetologists to discuss cases of uncertainty, could form the basis of a retinopathy screening service that accurately identifies and categorises retinopathy and does not miss sight-threatening cases.