Abstract
Purpose:
To validate the ability of vision technicians (VTs) in screening potentially blinding eye diseases and to find the interobserver agreement (with reference to ophthalmologist) for referral to higher level centers in a vision centre (VC) based model in rural South India.
Methods:
A random sample of subjects at the 7 VCs between July to August 2010 were simultaneously examined in masked manner by the 7 VTs and an ophthalmologist. The screening protocol included vision testing, refraction, slit-lamp biomicroscopy, applanation tonometry and undilated direct ophthalmoscopy examination. Inter-observer agreement and diagnostic accuracy parameters for referral pattern and screening of eye diseases were assessed using weighted kappa statistics. The diagnosis by the ophthalmologist was considered as the reference standard.
Results:
The subjects (n=279) with a mean age of 32.9+21.8 years were enrolled and examined at the 7 VCs. Agreement for diagnosis of ocular pathology was 0.82 (95% CI; 0.8-0.83). There was good agreement for the diagnosis of cataract (0.97; 95% CI, 0.93-1), refractive error (0.98; 95% CI, 0.96-1), corneal pathology (1.0; 95% CI, 1.0-1.0) and other anterior segment disorders (0.95; 95% CI, 0.9-1). The agreement was moderate to fair for the diagnosis of glaucoma and glaucoma suspects (0.43; 95% CI, 0.28-6) and retinal pathologies (0.39; 95% CI, 0.14-0.63). The sensitivity for diagnosis of anterior segment pathologies was good (94.6 to 100%); glaucoma and glaucoma suspects (35.6%; 95% CI, 21.9-51.2%) and retinal pathologies (26.3%; 95% CI, 9.2-51.2%) ranged from fair to moderate, while specificity for ocular pathologies (including refractive error) was between 98.2 to 100%. The overall kappa (κ) agreement for referral between the VTs and ophthalmologist was 0.79 (95% CI; 0.74-0.84).
Conclusions:
While there was good agreement between the VTs and ophthalmologist for referral and detection of anterior segment disorders, it was only was moderate to fair for glaucoma, glaucoma suspect and retinal pathologies.
Keywords: 460 clinical (human) or epidemiologic studies: health care delivery/economics/manpower