Purchase this article with an account.
Taras V Litvin, George H Bresnick, Glen Y Ozawa, Kuniyoshi Kanai, Jorge A Cuadros; Screening test characteristics of a novel hand-held ERG device in detection of referable diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 2016;57(12):3592.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the screening test characteristics of the photopic 30Hz flicker ERG measurements using a novel hand-held ERG device in detection of referable diabetic retinopathy, in a cohort of adults with type 1 and 2 diabetes mellitus.
Subjects were recruited at Eastmont Wellness Center, Oakland, CA. All subjects underwent a comprehensive ocular examination. Photopic 30Hz flicker ERG measurements were obtained from each eye using the RETeval hand-held ERG (LKC Technologies, Inc.). Fundus photographs were obtained and graded using the Canon CR-DGI fundus camera (Canon USA, Inc.), following the EyePACS protocol and grading system, a modified ETDRS protocol. Referable retinopathy was defined as severe non-proliferative or worse or clinically significant macular edema.
214 ethnically diverse patients were enrolled. The analysis is based on the 190 patients without pre-specified retinal pathology and with both RETeval device results and gradable photos available. 13 (6.8%) cases of referable retinopathy were identified. The mean age in the sample was 55±11 and 58% were females. Multivariable logistic regression analysis revealed that ERG Score, a measure derived by the RETeval device, and implicit time (IT) were significantly associated with referable retinopathy (p=0.006 and 0.002, respectively). One unit increase in Score was associated with the odds ratio (OR) of referable retinopathy equal to 1.5 (95%CI: 1.15-2.14). The sensitivity of Score for the detection of referable retinopathy was 76.9% (95% CI: 54%-99%). The specificity was 54% (95% CI: 47%-62). One millisecond delay in IT was associated with the OR of referable retinopathy equal to 1.8 (95%CI: 1.25-2.78). The optimal IT for classifying referable retinopathy was 28.15ms. The corresponding sensitivity and specificity were 92.3% (95% CI: 68.5% -100%) and 52.5% (95% CI: 37.4% - 63%), respectively.
An increase in Score and delay in IT showed increase in the odds of being diagnosed with referable retinopathy. RETeval Score had worse performance than reported elsewhere. This performance difference may be in part due to a difference in the gold standard, where the earlier work used 7-field stereo ETDRS photographs, and may also be in part due to a difference in the distribution of retinopathy and other clinical characteristics of the tested population.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only