June 2023
Volume 64, Issue 8
Open Access
ARVO Annual Meeting Abstract  |   June 2023
A Bayesian longitudinal hierarchical model for estimation of central visual field rates of change in glaucoma
Author Affiliations & Notes
  • Sajad Besharati
    Department of Ophthalmology, Jules Stein Eye Institute, Los Angeles, California, United States
  • Erica Su
    Department of Biostatistics, University of California Los Angeles, Los Angeles, California, United States
  • Vahid Mohammadzadeh
    Department of Ophthalmology, Jules Stein Eye Institute, Los Angeles, California, United States
  • Massood Mohammadi
    Department of Ophthalmology, Jules Stein Eye Institute, Los Angeles, California, United States
  • Joseph Caprioli
    Department of Ophthalmology, Jules Stein Eye Institute, Los Angeles, California, United States
  • Robert E Weiss
    Department of Biostatistics, University of California Los Angeles, Los Angeles, California, United States
  • Kouros Nouri-Mahdavi
    Department of Ophthalmology, Jules Stein Eye Institute, Los Angeles, California, United States
  • Footnotes
    Commercial Relationships   Sajad Besharati None; Erica Su None; Vahid Mohammadzadeh None; Massood Mohammadi None; Joseph Caprioli Research to Prevent Blindness, Payden Glaucoma Research Fund, the Simms/Mann Family Foundation., Code F (Financial Support); Robert Weiss None; Kouros Nouri-Mahdavi NIH R01 (EY029792); Departmental grant from Research to Prevent Blindness; Heidelberg Engineering, Code F (Financial Support)
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science June 2023, Vol.64, 5510. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sajad Besharati, Erica Su, Vahid Mohammadzadeh, Massood Mohammadi, Joseph Caprioli, Robert E Weiss, Kouros Nouri-Mahdavi; A Bayesian longitudinal hierarchical model for estimation of central visual field rates of change in glaucoma. Invest. Ophthalmol. Vis. Sci. 2023;64(8):5510.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : Efficient and accurate estimation of structural or functional rates of change benefits from proper modeling of such hierarchical longitudinal data. We have recently developed hierarchical longitudinal Bayesian models for modeling macular structural rates of change in glaucomatous eyes with moderate to advanced disease. We present a model version adapted for modeling local and global visual field rates of change in the same cohort of patients.

Methods : 124 eyes of 124 glaucoma patients with central or moderate to advanced field loss at baseline
were included. A hierarchical longitudinal random effects model was developed that accommodates
censoring of VF data at 0 dB and has correlated random subject residual variances, baseline levels, and
slopes. Population and subject rates of change were estimated in each of 68 test locations. The correlation
of intercepts and slopes at each individual test location with all the 67 other test locations were estimated.

Results : The average (SD) baseline 10-2 visual field mean deviation was –8.7 (5.6) dB. The patients were
followed for 4.6 (0.8) years and had a median (range) of 9 (4-13) visual field tests. Population slopes were
significant and negative in all but one locations, with posterior means ranging from –0.37 to –0.05
dB/year. Slopes were faster in superior locations compared to the inferior hemifield. The global mean
slope (95% credible interval) was –0.21 (–0.24, –0.18) dB/year. Figure 1 displays the correlation of the
slope at each location with all the other slopes (67 test locations) across the central 10-2 visual field.
Visual field slopes were less correlated with those of the adjacent locations across the temporal raphe.
There was a tendency towards faster slopes with decreasing threshold sensitivity (Figure 2).

Conclusions : Our newly developed Bayesian hierarchical longitudinal model addresses many unique
challenges related to modeling of longitudinal visual field data. This model is able to estimate the spatial
correlation of adjacent test locations as a preliminary step to analyzing and decomposing the relationship
between the variances of the hyperparameters in order to account such spatial correlations.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

 

 

×
×

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.

×