Abstract
Purpose :
There is a high demand for accurate and cost-effective diabetic retinopathy (DR) screening programs. We conducted a meta-analysis investigating how factors such as mydriasis, field number and field size affect the accuracy of photographic diabetic retinopathy screening, when compared to ETDRS or dilated fundus examination by an ophthalmologist.
Methods :
A database search was conducted through MEDLINE, WoS, CINAHL, The Cochrane Library, Rural and Remote Health Database (RURAL) between the period January 1990 to December 2015. Data was extracted from 58 studies by protocol and methodological quality was assessed using a standard checklist. Levels of retinopathy were separated into any DR (ADR) and sight-threatening DR (STDR). Mixed effects meta-analysis of sensitivity and specificity were performed according to the number of image fields assessed and the level of DR required for a positive result. Wide-angle (≥ 200) imaging techniques were assessed separately.
Results :
For ADR and non-wide-angle images, there was a trend of increasing sensitivity with the number of fields captured (0.80 for single field, 0.87 for 2-3 fields, 0.98 for 4-7 fields). Specificity remained fairly stable across number of fields (ranging from 0.91 to 0.94). The pooled estimate of positive likelihood ratio was 4.8 for wide angle images and greater than 9.2 for all other categories, suggesting a positive-screen result is highly indicative of disease. For STDR, a trend of increasing sensitivity with increasing number of fields was also seen (0.73 for single field, 0.85 for 2 fields, 0.73 for 3 fields, 0.93 for 4-7 fields). For each number of fields assessed, specificity for STDR was high (0.97 to 0.98) with high positive likelihood ratios. For wide-angle images the specificity was lower than estimated for standard field sizes (0.8 any DR and 0.9 sight-threatening DR). Meta-regression for the detection of ADR using single or 2 fields less than 200° revealed no differences in sensitivity or specificity across study location, year, mydriatic status and medium of photography (Digital vs. Film). Mydriasis was associated with improved sensitivity for detection of STDR using 2 image fields.
Conclusions :
The sensitivity and specificity of photographic screening for ADR and STDR was positively associated with number of photographic fields. There was no evidence of improved sensitivity or specificity following pupil dilation for the detection of ADR.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.