N Ghazali1, T Aslam1,2, D B Henson1,2. 1. Institute of Human Development, Faculty of Medical and Human Development, University of Manchester, Manchester, UK. 2. Manchester Royal Eye Hospital, Central Manchester University Hospitals NHS Foundation Trust, Manchester Academic Health Science Centre, Manchester, UK.
Abstract
PURPOSE: To describe and measure the discriminatory performance of three new superior-inferior asymmetry indices for detecting primary open angle glaucoma (POAG) and to compare these with the glaucoma hemifield test (GHT). METHODS: In all, 412 control and 247 POAG eyes were selected from a visual field database of patients attending the Manchester Royal Eye Hospital. Age-adjusted defect asymmetries were calculated for each of the 22 vertically mirrored test point pairs used in the GHT. The three new indices, hemifield mean difference (HMD) and hemifield standard deviation (HSD) of the asymmetry values along with the number of test pairs (NP) falling outside the 85% probability limits of the control population, were calculated. ROC curves of the indices and GHT were constructed. Agreement between the indices was explored with a proportional Venn diagram and 3 × 3 contingency tables. Cases of disagreement between the indices were reviewed. RESULTS: The area under the ROC curves were HMD=0.745 (95% confidence interval (CI) 0.705-0.786), HSD=0.864 (95% CI 0.833-0.894), NP=0.863 (95% CI 0.832-0.893) and GHT=0.792 (95% CI 0.754-0.829). The Venn diagram and contingency tables highlighted the good agreement between HSD, NP and GHT. Agreement was 78% (HSD vs. GHT) and 82% (NP vs. GHT) in the control sample and 70% (HSD vs. GHT) and 71% (NP vs. GHT) in the POAG sample. Five cases are presented where disagreement existed between the indices. CONCLUSIONS: The new HSD and NP asymmetry indices perform better than GHT in differentiating between normal and POAG eyes in this data set. GHT can fail to detect significant asymmetry, detected by HSD and NP, when an early defect crosses sector boundaries.
PURPOSE: To describe and measure the discriminatory performance of three new superior-inferior asymmetry indices for detecting primary open angle glaucoma (POAG) and to compare these with the glaucoma hemifield test (GHT). METHODS: In all, 412 control and 247 POAG eyes were selected from a visual field database of patients attending the Manchester Royal Eye Hospital. Age-adjusted defect asymmetries were calculated for each of the 22 vertically mirrored test point pairs used in the GHT. The three new indices, hemifield mean difference (HMD) and hemifield standard deviation (HSD) of the asymmetry values along with the number of test pairs (NP) falling outside the 85% probability limits of the control population, were calculated. ROC curves of the indices and GHT were constructed. Agreement between the indices was explored with a proportional Venn diagram and 3 × 3 contingency tables. Cases of disagreement between the indices were reviewed. RESULTS: The area under the ROC curves were HMD=0.745 (95% confidence interval (CI) 0.705-0.786), HSD=0.864 (95% CI 0.833-0.894), NP=0.863 (95% CI 0.832-0.893) and GHT=0.792 (95% CI 0.754-0.829). The Venn diagram and contingency tables highlighted the good agreement between HSD, NP and GHT. Agreement was 78% (HSD vs. GHT) and 82% (NP vs. GHT) in the control sample and 70% (HSD vs. GHT) and 71% (NP vs. GHT) in the POAG sample. Five cases are presented where disagreement existed between the indices. CONCLUSIONS: The new HSD and NP asymmetry indices perform better than GHT in differentiating between normal and POAG eyes in this data set. GHT can fail to detect significant asymmetry, detected by HSD and NP, when an early defect crosses sector boundaries.