Abstract
Purpose :
To report diabetic tele-retinopathy screening results from a screening program in the San Francisco Health Network, a health care system providing care for a large urban and multi-ethnic patient population. Ultrawide field (UWF) cameras were deployed on a mobile van unit and in primary care clinics. Predictors for presence of diabetic retinopathy (DR) and proliferative diabetic retinopathy (PDR) were also analyzed.
Methods :
Cross sectional study. Adult diabetics screened from 3/2/2017 – 7/17/2018 were included. Patients were excluded if fundus photos were uninterpretable by the reading center. Patient demographics, comorbidities, medications, social history, hemoglobin A1C (HbA1c), body mass index, evidence of end-organ damage, ocular comorbidities and fundus tele-retinopathy interpretations were recorded. Univariate logistic regression was utilized to estimate the odds ratio for presence or absence of DR and presence or absence of PDR for all baseline characteristics. Multivariate logistic regression was performed adjusting for age, gender, and factors associated with presence of DR or PDR in univariate analysis.
Results :
Median age was 60 years. 2,788 patients were screened, with 59 (2%) excluded due to un-gradable fundus images from cataract or media opacity. Of the remaining 2,729 patients, 1,993 (73%) had no DR. Of the 736 (27%) that had DR, 34 (5%) had PDR. Of the 702 (95%) with non-proliferative diabetic retinopathy (NPDR), 534 (76%) had mild NPDR, 147 (21%) had moderate NPDR and 19 (2.7%) had severe NPDR. The predictors associated with presence of any DR on multivariate analysis were diabetes duration, hemoglobin A1C level, insulin use, and presence of end-organ damage. The single predictor associated with presence of PDR on multivariate analysis was duration of DM. The use of the camera on an eye screening van was straightforward, with high image quality comparable to the stationary UWF cameras (Chi squared, P<0.005).
Conclusions :
In this multi-ethnic, urban patient population, diabetes duration, HbA1C level, insulin use, and presence of end-organ damage were associated with presence of DR, and diabetes duration was associated with presence of PDR. The use of a mobile UFW device expanded the catchment area and is promising in expanding access to screenings beyond the traditional clinical setting.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.