Identifiable Historic and Observable Factors May Predict Progression to Exfoliation Glaucoma in Newly Diagnosed Exfoliation Patients.

Abstract

Objective: To identify clinical factors associated with conversion to exfoliation glaucoma (XFG) in exfoliation syndrome (XFS) patients who are most at risk of progression to XFG within 3 years for increased surveillance and early preventive interventions.

Design: A retrospective patient cohort study design was employed.

Subjects: A source population of XFS patients ≥ 50 years was identified from electronic medical records in the Utah Population Database. From this, 487 study patients with one or more dilated eye examinations before chart-confirmed XFS onset in 2011 or later and ≥ 3 years of subsequent eye examinations were selected for study.

Methods: We implemented binomial linear mixed models with L1-penalized estimation to select variables associated with conversion. Models included a random intercept to account for within-patient correlation for eye-level data. Candidate demographic, lifestyle, systemic, and ocular comorbidities data were obtained, and diagnoses were categorized as binary (history or no history). These potential factors between conversion and nonconversion patients were used in model selection of variables jointly predictive of conversion. Odds ratios and confidence intervals were calculated using the link logit.

Main outcome measures: To determine the main outcome of conversion to XFG following an index diagnosis of XFS compared with nonconversion within 3 years, clinical records of each subject's left and right eyes were assessed to confirm XFS and date of onset and date of XFG onset, if conversion occurred. Clinical measurements (e.g., intraocular pressure [IOP], cup-to-disc ratio, provider notes, and IOP-lowering procedures and medications) were used to corroborate conversion status.

Results: Eighteen variables jointly predicted XFG conversion within 3 years correctly in 71% of patient eyes. The odds of conversion were the highest for exudative age-related macular degeneration (AMD), 2.3-fold (P = 0.004). Other predictive variables included nonexudative AMD (P = 0.05), primary open angle glaucoma (P < 0.001), obstructive sleep apnea (P = 0.03), and ocular hypertension (P = 0.003) diagnosed before XFS onset.

Conclusions: We determined a set of clinically relevant factors that predicted which newly diagnosed XFS patients progressed to XFG within 3 years. A planned validation will independently confirm if these prognostic indicators hold promise in other settings.

Financial disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.