Risk Factors for Optic Disc Hemorrhage in the Low-Pressure Glaucoma Treatment Study

PURPOSE: To investigate risk factors for disc hemorrhage detection in the Low-Pressure Glaucoma Treatment Study. DESIGN: Cohort of a randomized, double-masked, multicenter clinical trial. METHODS: Low-Pressure Glaucoma Treatment Study patients with at least 16 months of follow-up were included. Exclusion criteria included untreated intraocular pressure (IOP) of more than 21 mm Hg, visual field mean deviation worse than - 16 dB, or contraindications to study medications. Patients were randomized to topical treatment with timolol 0.5% or brimonidine 0.2%. Stereophotographs were reviewed independently by 2 masked graders searching for disc hemorrhages. The main outcomes investigated were the detection of disc hemorrhage at any time during follow-up and their recurrence. Ocular and systemic risk factors for disc hemorrhage detection were analyzed using the Cox proportional hazards model and were tested further for independence in a multivariate model. RESULTS: Two hundred fifty-three eyes of 127 subjects (mean age, 64.7 +/- 10.9 years; women, 58%; European ancestry, 71%) followed up for an average standard deviation of 40.6 +/- 12 months were included. In the multivariate analysis, history of migraine (hazard ratio [HR], 5.737; P = .012), narrower neuroretinal rim width at baseline (HR, 2.91; P = .048), use of systemic beta-blockers (HR, 5.585; P = .036), low mean systolic blood pressure (HR, 1.06;P = .02), and low mean arterial ocular perfusion pressure during follow-up (HR, 1.172; P = .007) were significant and independent risk factors for disc hemorrhage detection. Treatment randomization was not associated with either the occurrence or recurrence of disc hemorrhages. CONCLUSIONS: In this cohort of Low-Pressure Glaucoma Treatment Study patients, migraine, baseline narrower neuroretinal rim width, low systolic blood pressure and mean arterial ocular perfusion pressure, and use of systemic beta-blockers were risk factors for disc hemorrhage detection. Randomization assignment did not influence the frequency of disc hemorrhage detection. (C) 2014 by Elsevier Inc. All rights reserved.)