May 2005
Volume 46, Issue 13
ARVO Annual Meeting Abstract  |   May 2005
Comparison of Methods to Predict Visual Field Progression in Glaucoma
Author Affiliations & Notes
  • K. Nouri–Mahdavi
    Glaucoma Division, Jules Stein Eye Institute, Los Angeles, CA
  • D. Hoffman
    Glaucoma Division, Jules Stein Eye Institute, Los Angeles, CA
  • D. Gaasterland
    University Ophthalmology Consultants of Washington, Washington, DC
  • J. Caprioli
    Glaucoma Division, Jules Stein Eye Institute, Los Angeles, CA
  • Footnotes
    Commercial Relationships  K. Nouri–Mahdavi, None; D. Hoffman, None; D. Gaasterland, None; J. Caprioli, None.
  • Footnotes
    Support  Fight for Sight Grant–in–Aid (KNM), NIH R01 EY12738, and RPB
Investigative Ophthalmology & Visual Science May 2005, Vol.46, 3729. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      K. Nouri–Mahdavi, D. Hoffman, D. Gaasterland, J. Caprioli; Comparison of Methods to Predict Visual Field Progression in Glaucoma . Invest. Ophthalmol. Vis. Sci. 2005;46(13):3729.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

The purpose of this investigation is to compare the performance of 3 methods: PLR, the Glaucoma Change Probability Analysis (GCPA), and the AGIS criteria, to predict the visual field status at 8 years from VF data available at 4 years of follow–up.


This study included 183 eyes (161 patients) from the AGIS followed for a minimum of 8 years with at least 6 VF available during the first 4 years and at least 6 VF during the second four years. The performance of GCPA (based on criteria for definite progression in EMGT) was compared to that of PLR with stringent criteria at 4 and 8 years, and to AGIS criteria.


Based on change from baseline in at least one of the assessment methods, progression occurred in 71 (39%) eyes at 8 years. However, the 3 methods agreed on progression in 22 of the 71 eyes (31%). 53 of 183 eyes (29%) progressed based on PLR criteria, 55 (30%) eyes progressed based on GCPA criteria, and 31 (17%) progressed based on AGIS criteria. The kappa values for pairwise agreement of the 3 methods were: 0.41 (95% CI: 0.25–0.57) for agreement of AGIS and GCPA, 0.52 (95% CI: 0.36–0.68) for AGIS and PLR, and 0.66 (0.54–0.78) for GCPA and PLR. The highest agreement at 4 years was also observed between PLR and GCPA (kappa = 0.47; 95% CI: 0.23–0.71). GCPA at 4 years detected 21 of 55 eyes with GCPA progression at 8 years (38%, 95% CI: 31–45%). PLR at 4 years detected 14 of 53 eyes with PLR progression at 8 years (26%, 95% CI: 20–32%). The AGIS scoring system at 4 years detected 9 of 31 eyes with AGIS progression at 8 years (29%, 95% CI: 22–36%). The false positive rates were 0.8%, 1.5%, and 2.6%, respectively.


Our results suggest that GCPA detects visual field progression earlier compared with PLR; AGIS criteria are intermediate. AGIS criteria were most conservative and best agreement was observed between PLR and GCPA. All methods were highly specific when the results of the same methods at 8 years were compared to those of 4 years.



Keywords: visual fields • clinical (human) or epidemiologic studies: outcomes/complications 

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.