PURPOSE: To examine the relationships between baseline risk factors and sustained decrease of visual field (SDVF) and sustained decrease of visual acuity (SDVA).

DESIGN: Cohort study of participants in the Advanced Glaucoma Intervention Study (AGIS).

METHODS: This multicenter study enrolled patients between 1988 and 1992 and followed them until 2001; 789 eyes of 591 patients with advanced glaucoma were randomly assigned to one of two surgical sequences, argon laser trabeculoplasty (ALT)-trabeculectomy-trabeculectomy (ATT) or trabeculectomy-ALT-trabeculectomy (TAT). This report is based on data from 747 eyes. Eyes were offered the next intervention in the sequence upon failure of the previous intervention. Failure was based on recurrent intraocular pressure elevation, visual field defect, and disk rim criteria. Study visits occurred every 6 months; potential follow-up ranged from 8 to 13 years. For each intervention sequence, Cox multiple regression analyses were used to examine the baseline characteristics for association with two vision outcomes: SDVF and SDVA. The magnitude of the association is measured by the hazard ratio (HR), where HR for binary variables is the relative change in the hazard (or risk) of the outcome in eyes with the factor divided by the hazard in eyes without the factor, and HR for continuous variables is the relative change in the hazard (or risk) of the outcome in eyes with a unit increase in the factor.

RESULTS: Characteristics associated with increased SDVF risk in the ATT sequence are: less baseline visual field defect (hazard ratio [HR] = 0.86, P < .001, 95% CI = 0.82–0.90), male gender (HR = 2.23, P < .001, 1.54–3.23), and worse baseline visual acuity (HR = 0.96, P = .001, 0.94–0.98); in the TAT sequence: less baseline visual field defect (HR = 0.93, P = .001, 0.89–0.97) and diabetes (HR = 1.87, P = .007, 1.18–2.97). Characteristics associated with increased SDVA risk in both treatment sequences are better baseline acuity (ATT: HR = 1.05, P < .001, 1.02–1.09; TAT: HR = 1.06, P < .001, 1.03–1.08), older age (ATT: HR = 1.05, P = .001, 1.02–1.08; TAT: HR = 1.04, P = .002, 1.01–1.06), and less formal education (ATT: HR = 1.92, P = .001, 1.29–2.88; TAT: HR = 1.77, P = .002, 1.22–2.54).

CONCLUSIONS: For SDVF, risk factors were better baseline visual field in both treatment sequences, male gender, and worse baseline visual acuity in the ATT sequence, and diabetes in the TAT sequence. For SDVA, risk factors in both treatment sequences were better baseline visual acuity, older age, and less formal education.