ADA – New Screening Methods Work for Diabetic Retinopathy

These screening approaches appeared to retain good specificity and sensitivity for diabetic retinopathy detection while reducing the burden on ophthalmologists and patients in two studies reported at the American Diabetes Association meeting.
These kinds of studies are important to improving the rates of annual screening recommended for all diabetic patients by the ADA, particularly in underserved areas, commented Sue Kirkman, M.D., ADA vice-president of clinical affairs.

In the U.S., the average retinal screening rate is only 65%, noted Athena Philis-Tsimikas, M.D., of the Scripps Whittier Diabetes Institute in San Diego, CA, and colleagues.

Since that rate falls even lower in underserved communities, her group partnered with community health centers to bring a mobile medical unit equipped with a nonmydriatic digital retinal camera to those areas in San Diego County.

Over an 18-month period, 1,229 people with diabetes and no or minimal insurance were screened.

A trained technician took retinal pictures, determined the presence of significant retinopathy using EyePacs imaging software, and provided immediate referral if it was present. A retinal specialist independently read all the images to determine accuracy of the screening. Comparing them, the technician’s screening had an overall accuracy of 92% with a specificity of 96% and sensitivity of 86%.

The negative predictive value was 79%; the positive predictive value was 97%.

Dr. Philis-Tsimikas stated that, the researchers did not calculate number needed to screen, but diagnosed diabetic retinopathy in 183 of 1,229 screened individuals. With this screening strategy, the technician could do preliminary screening, with only positive screens sent to a retinal specialist for review to maximize use of the specialist’s time and expertise. Patients also benefited from accurate, sameday evaluation without long wait times as well as an opportunity for patient education.
The accuracy of another strategy to increase efficiency and resource utilization was examined in a second study by Jyothis George, M.B.B.S., of the Queen’s Medical Research Institute in Edinburgh, Scotland, and colleagues. Dilated seven-field retinal photography is the gold standard but “too resource intensive to be used for screening,” they noted.

Dr. George mentioned that Single-field photography — which encompasses only the central 45º of the retina most vital for vision-threatening complications — without dilation is more popular for screening. His group compared the two approaches in a subanalysis of the Edinburgh Type 2 Diabetes Study that included 751 asymptomatic individuals screened with dilated, seven-field methods who also had routine single-field retinal photography within a year. Despite small numbers, the researchers found 92.3% sensitivity (95% CI 73 to 98) for diabetic retinopathy detection and referral for further investigation in the patients with at least moderate diabetic retinopathy (24 of 26 cases). Specificity was even higher at 99.8% (99.1-99.9) for single-field screening.

There was one false positive with single-field screening — a case of venous occlusion wrongly classified as referable diabetic retinopathy.

However, the performance of single-field methods in the overall cohort was not as good:

• 65.4% sensitivity (95% CI 59.1 to 71.3)
• 86.5% specificity (95% CI 83 to 89.3)
• 12.8% of those without diabetic retinopathy were labeled as mild cases.
• 36.7% with mild diabetic retinopathy were missed.

But the researchers said, “This is clinically not significant as our screening strategy is to recall all these patients for yearly single-field nonmydriatic retinal screening.”

They concluded that their single-field method would be accurate “when used in conjunction with integrated

George JT, et al “Sensitivity and specificity of single field non-mydriatic retinal photography in screening for diabetic retinopathy: The Edinburgh Type 2 Diabetes Study” ADA 2009; Abstract P880.