Screening for diabetic retinopathy

Combined modalities seem to provide the best option

Diabetic retinopathy is an important cause of blindness,1 and the British Diabetic Association has recently begun a nationwide campaign to increase awareness of the problem. Much blindness due to diabetes is preventable: if sight threatening retinopathy is detected in time then laser treatment can greatly reduce the progression to blindness.2 To detect those patients who require treatment the traditional recommendation is that every one to two years all diabetic patients should be screened either by ophthalmoscopy through pharmacologically dilated pupils or by retinal photography.3

But concern is increasing that this advice is insufficient and the Department of Health and Social Security in the late 1980s mounted the largest British study of the subject, which recruited 3318 patients to evaluate possible screening modalities.4 This study showed that ophthalmoscopy through dilated pupils (whether by general practitioners, optometrists, or hospital physicians) and retinal photography through undilated pupils both missed between one third and two thirds of cases of sight threatening diabetic retinopathy. The study concluded, “in a setting close to routine screening, the sensitivities of all screening methods are poor …, the routine use of any of these single screening methods will fail to detect a large proportion of cases of sight threatening diabetic retinopathy.”4

Similar results have been found in other studies.5 These studies, and a study showing that ophthalmologists and diabetologists may miss appreciable amounts of sight threatening diabetic retinopathy if allowed only dilated ophthalmoscopy in routine screening,6 suggest that ophthalmoscopy is unreliable as a means of screening, no matter who does it.

The Department of Health's study that suggested that 45° field retinal photography was as unreliable as ophthalmoscopy for screening for retinopathy used retinal photography through undilated pupils.4 This was because the 45° field cameras were being marketed at the time as “non-mydriatic.” Diabetic patients tend to have abnormally small pupils as a manifestation of autonomic neuropathy,7 8 and small pupils lead to poor quality photographs. As the complications of diabetes tend to cluster in the same patients,9 the patients with autonomic neuropathy and abnormally small pupils are more likely to be the ones who have diabetic retinopathy.

Studies of “non-mydriatic” retinal photography have shown that when it is used in diabetic patients poor quality photographs occur in about a quarter.4 10 11 Studies using the same cameras through pharmacologically dilated pupils have shown that the photographic quality is greatly improved.11 12 Had retinal photography through pharmacologically dilated pupils been included in the Department of Health's study much less sight threatening diabetic retinopathy might have been missed. Nevertheless, the possibility of sight threatening diabetic retinopathy beyond the field of the standard 45° photograph exists.13

Another large study of 2159 patients showed that both ophthalmoscopy and retinal photography miss considerable amounts of sight threatening diabetic retinopathy detected by the other technique.14 A report has recently been published of combined ophthalmoscopy and retinal photography, both through dilated pupils, performed by a trained technician.15 Though this study lacked full assessment by an ophthalmologist of all 1050 diabetic patients, close agreement existed between the assessment of the technician and the ophthalmologist's assessment based on the photograph; the technique seemed to have high specificity: among 45 patients suspected of having sight threatening diabetic eye disease on the basis of screening, only one referral was thought inappropriate.

Combined dilated retinal photography and ophthalmoscopy have also been adopted in west Birmingham, where the screening is done by specialist optometrists highly experienced in diabetic retinopathy. The optometrist can call on a diabetologist to discuss cases of doubt while the patient is still present with eyes dilated. An audit of this system presented recently to the British Diabetic Association suggested that the combined modalities in this setting have abolished the problem of appreciable amounts of sight threatening diabetic retinopathy being missed by ophthalmoscopy alone and have also generated a sensitive system for detecting diabetic retinopathy in general.16 Other centres are adopting dual screening systems and are finding them similarly successful.17

Those who plan screening programmes for diabetic retinopathy should be aware that either ophthalmoscopy through dilated pupils alone or retinal photography through undilated pupils will miss between one third and two thirds of cases of sight threatening diabetic retinopathy, even if performed by an experienced ophthalmoscopist. They should also bear in mind a patient who claimed that she had gone blind because her retinopathy had not been adequately detected or treated and who recently settled her case out of court, receiving pounds sterling225 000 from her health authority.18

The cost of litigation may dwarf into insignificance the cost of providing screening programmes combining the mutual benefits of ophthalmoscopy and retinal photography through dilated pupils with screening being administered by experienced personnel.