Agreement in Measurement of Optic Cup-to-Disc Ratio with Stereo Biomicroscope Funduscopy and Digital Image Analysis: Results from the Nigeria National Blindness and Visual Impairment Survey. (2nd January 2017)

• Record Type: Journal Article
• Authors: Kyari, Fatima; Gilbert, Clare
• Is Part Of: Ophthalmic epidemiology. Volume 24:Number 1(2017)
• Page Start: 57
• Page End: 62
• Publication Date: 2017-01-02
• Subjects: Bland-Altman plot -- glaucoma -- limits of agreement -- Nigeria -- vertical cup-to-disc ratio (VCDR); Blindness -- Prevention -- Periodicals; Eye -- Diseases -- Epidemiology -- Periodicals; Ophthalmology -- Periodicals; 614.5997
• DOI: 10.1080/09286586.2016.1254806
• Languages: English
• ISSNs: 0928-6586

Abstract

ABSTRACT: Purpose : To determine agreement in estimations of vertical cup-to-disc ratio (VCDR) between clinical stereo-biomicroscopic funduscopy and digital fundus image analysis. Methods : Systematic sampling of 1-in-7 from a sample of 13, 591 participants aged ≥40 years gave a subsample who were examined in detail. VCDR was estimated clinically by 60 diopter aspheric lens biomicroscopic funduscopy (c-VCDR) and by digital fundus images (i-VCDR) graded at the Moorfields Eye Hospital Reading Centre. Spearman's correlation coefficient, paired t-test and the Bland-Altman method to assess limits of agreement (LOA) between the two methods were applied. Results : Of 1759 participants in the subsample, 848 participants (48%) with normal frequency doubling technology (FDT) visual fields and data for i-VCDR and c-VCDR in both eyes ( n = 1696 eyes) were included in the analysis. By absolute difference of VCDR values for each eye between the two methods, 1585 eyes (94%) differed by ≤0.2. Mean i-VCDR was 0.381 (standard deviation, SD 0.156), and mean c-VCDR 0.321 (SD 0.145). i-VCDRs were significantly larger by a mean difference of 0.061 (SD 0.121; 95% confidence interval, CI, 0.055–0.066; p < 0.001). The 95% LOA assessed by the Bland-Altman method were lower limit −0.182 (95% CI −0.192 to −0.172) and upper limit 0.303 (95% CI 0.293–0.313). The 95% LOA intervals narrowed with higher VCDR. Conclusion : Digital image analysis and clinical assessment are two distinct methods to measure VCDR; with larger i-VCDRs in this survey. Applying i-VCDR cut-off values to c-VCDR measurements in the Nigeria Blindness Survey might have underestimated glaucoma prevalence. It is recommended that all participants in glaucoma surveys have VCDR by digital image measurement.