Abstract
Purpose:
The burden of diabetes and diabetic retinopathy (DR) is increasing worldwide. Timely screening is an important strategy to prevent needless blindness from DR. One reasonable approach to high-quality DR screening is the use of a non-mydriatic fundus camera However, most previously tested cameras are either tabletop models limiting their portability and applicability in rural outreach settings, or have poor resolution. The objective of this study is to evaluate the sensitivity and specificity of a relatively inexpensive handheld 45° non-mydriatic fundus camera (Smartscope, Optomed, Finland) to aid in the remote diagnosis of vision-threatening diabetic retinopathy (VTDR).
Methods:
Subjects were recruited from a convenience sample of diabetic and non-diabetic patients accessing the vitreoretinal service at the Aravind Eye Hospital, Pondicherry, India. Three undilated images (macula, nasal and superotemporal fields) were obtained from each eye with the Smartscope. The subject then underwent complete dilated fundus examination by a retina specialist as the gold standard followed by dilated fundus imaging with the Smartscope and a standard tabletop fundus camera (Topcon, Japan). All the images were de-identified and graded remotely by two masked graders (retina specialists) using the NHS guidelines for DR grading. VTDR was defined as the presence of R2 level disease (≥ severe non-proliferative DR) and/or diabetic macular edema.
Results:
We report preliminary results on 134 eyes of 81 subjects, of whom 78% had diabetes for an average duration of 6 years and 37% had VTDR on clinical exam. Both graders reported ≥1 gradable image in 88% of eyes using the undilated Smartscope and in 95% eyes using either the dilated Smartscope or the Topcon. Graders 1 and 2 reported a sensitivity of 93% and 74% and a specificity of 82% and 97% respectively in identifying VTDR (k=0.66) with the undilated Smartscope. Graders were relatively more concordant when reporting VTDR interpreting the dilated Smartscope (sensitivity=92%, 85% and specificity=81%, 96% respectively) (k=0.68) and Topcon images (sensitivity=96%, 94% and specificity=91%, 94% respectively) (k=0.89).
Conclusions:
The Smartscope offers reasonable quality non-mydriatic images and it shows initial promise in meeting British Diabetic Association standards of a sensitivity ≥80% to be considered a useful portable tool for screening for VTDR in community outreach programs.