Glaucoma is a heterogenous group of diseases characterised by cupping of the optic nerve head and visual-field damage.1 It is the most frequent cause of irreversible blindness worldwide.2 Interpretation of visual fields is an essential part of diagnosis, severity grading, and prognostication in glaucoma. However, clinical assessment of visual fields is unreliable due to subjectivity and ambiguity in guidance documentation.3 This leads to differential treatment of patients based on arbitrary factors which can lead to adverse outcomes. For instance, decisions to certify patients as visually impaired depend on the severity of patients’ visual field defects which are explicitly defined as a clinical decision (rather than being based on explicit objective criteria) in United Kingdom guidelines.4 Patients eligible for social support due to sight impairment are frequently unregistered as a consequence of such subjectivity.5–7 Glaucoma patients eligible on the basis of visual field defects are significantly more likely to miss out on registration than patients eligible on the basis of visual acuity, because of significant disagreement between ophthalmologists evaluating visual fields using idiosyncratic and subjective clinical criteria.6,8–10 Moreover, risk stratification of glaucoma patients is a priority when timely glaucoma care is challenged by increasing demand for services, as was highlighted during the COVID pandemic. Staging glaucomatous field defects is an important component of such risk stratification.11
Standardised rules for classifying glaucomatous field defects were proposed by Hodapp, Parish, and Anderson in 1993, who defined ‘early’, ‘moderate’, and ‘severe’ defects based on the mean deviation, global plot, and pattern deviation on Humphrey 24 − 2 test printouts.12 The Hodapp-Parish-Anderson (HPA) criteria have been used widely in research studies for their clarity and reproducibility, and HPA decisions align closer with glaucoma subspecialists than general ophthalmologists without specific expertise.9,13 However, use in regular clinical practice is limited due to the labour intensive requirement to evaluate multiple parameters presented on perimetry plots for every assessment. Accelerating incorporation of HPA criteria into clinical workflows could improve the accuracy, reliability, and fairness of visual field assessment for glaucoma patients. Here, GFDC (Glaucoma Field Defect Classifier), a web-application which automates grading based on HPA criteria without requiring patient-identifiable data to be inputted, is presented and validated.