Handheld Retinal Imaging Devices Show Utility for Diabetic Retinopathy Screening

Handheld retinal imaging devices may improve diabetic retinopathy (DR) screening in underserved communities, according to research published in the British Journal of Ophthalmology. Increasing the number of imaging fields may increase sensitivity and specificity in detecting referable DR, according to the report.

“The ever-increasing DR burden in resource-poor settings needs to be addressed innovatively,” the study authors explain. “DR screening with teleophthalmology using handheld retinal imaging devices is one such approach that can broaden the reach of [DR screening programs] in accessing a wider and often hard-to-reach population while increasing the identification of [referable DR].”

Researchers included patients with diabetes (N=116; mean age, 58.6 years; 58.6% women) in the single-site, cross-sectional, comparative validation study which compared agreement between 3 handheld retinal imaging devices and standard 7-field Early Treatment Diabetic Retinopathy Study (ETDRS) photography. Among the 3 devices, 2 provided 1-, 2-, and 5-field imaging and 1 provided 1- and 2-field imaging. Study participants underwent retinal imaging with all 4 devices and 5 trained graders evaluated the images.

The standard ETDRS photography assessment revealed 33.3% of participants had no DR, 10.4% had mild non proliferative DR (NPDR), 14.2% had moderate NPDR, 11.6% had severe NPDR, and 20.4% had proliferative DR.

Compared with the standard EDTRS grading, the images demonstrated moderate to substantial agreement in all fields. For 1-field imaging, the devices showed 59.6% (weighted k [kw], 0.77), 44.4% (kw, 0.54), and 42.2% (kw, 0.51) agreement, respectively. Increasing the number of fields resulted in greater agreement for the 2 handheld retinal imaging devices capable of 5-field imaging, but not for the 2-field device — 2-field imaging yielded 63.1% agreement (kw, 0.75), 49.3% (kw, 0.60), and 48.4% (kw, 0.59) among devices 1, 2, and 3, respectively, and 5-field imaging yielded agreement of 65.8 (kw, 0.75) and 60.0% (kw, 0.73) among the 2 devices with 5-field capability.

One 5-field handheld retinal imaging device had ungradable rates of 2.23% for 1-field, 1.79% for 2-field, and 0% for 5-field compared with 7.59% for 1-field, 4.02% for 2-field, and 3.56% for 5-field in the other device. The 2-field device had ungradable rates of 6.67% and 5.78% for 1-field and 2-field images, respectively.

The sensitivities and specificities ranged between 0.69-0.88 and 0.89-0.99 for one of the 5-field devices, between 0.67-0.88 and 0.82-0.92 for the other 5-field device, and between 0.78-0.91 and 0.89-0.97 for the 2-field handheld retinal imaging device, respectively.

“These data suggest that handheld retinal imaging performed with additional fields and certain specific systems is accurate enough for [DR screening programs] where their size, cost and ease of use would allow them to be widely deployed in underserved settings,” according to the researchers. “When using a [5-field] imaging protocol, the ungradable rate can be reduced to a minimum, as low as 0% with some systems, leading to further improvement in patient care.”

Study limitations include a single center design and ethnic homogeneity in the study sample, which may limit the globalization of these findings.

Some study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures.