Table 5.
Multivariate analysis using backward selectiona to assess risk factors for blindness (96 subjects, 102 eyes), Primary Open-Angle African-American Glaucoma Genetics study, United States.
| N | n (%) | Odds ratio (95% CI) | p-valueb | |
|---|---|---|---|---|
| Best-corrected visual acuity at diagnosis (Snellen) | 0.002 | |||
| 20/20 or better | 55 | 13 (23.6) | 1 (reference) | – |
| 20/25 to 20/40 | 25 | 11 (44.0) | 1.82 (0.44–7.53) | 0.41 |
| Worse than 20/40 | 22 | 19 (86.4) | 27.10 (4.21–174.00) | 0.0005 |
| Missed visits/year, n | 0.001 | |||
| 0–1 | 31 | 7 (22.6) | 1 (reference) | – |
| >1–2 | 36 | 10 (27.8) | 1.23 (0.26–5.88) | 0.79 |
| >2 | 26 | 18 (69.2) | 12.40 (2.32–66.30) | 0.003 |
| Unknownc | 9 | 8 (88.9) | 64.30 (12.40–335.00) | <0.0001 |
| Visits with elevated IOP (≥21 mmHg), % | 0.0002 | |||
| 0 | 32 | 4 (12.5) | 1 (reference) | – |
| >0–20 | 36 | 17 (47.2) | 25.30 (5.46–117.00) | <0.0001 |
| >20 | 34 | 22 (64.7) | 72.60 (15.30–344.00) | <0.0001 |
Initial model included diabetes, age at diagnosis, best-corrected visual acuity category at diagnosis, pre-treatment IOP, initial number of medications started, number of medications at study endpoint, number of glaucoma lasers during study period, number of treatment changes, number of missed appointments/year, percent of times non-adherence noted in chart, percent of visits with elevated IOP, average IOP throughout treatment, and IOP standard deviation.
From generalized linear model using generalized estimating equation to account for correlations from matching and paired eyes of a subject.
Unknown category represents patients whose follow-up predated records for missed appointments (prior to 2000) and from cases who presented blind.
CI, confidence interval; IOP, intraocular pressure.