June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Determining the Accuracy of Rates of RNFL Loss Measured by Peripapillary OCT
Author Affiliations & Notes
  • Chris Bradley
    Johns Hopkins Medicine Wilmer Eye Institute, Baltimore, Maryland, United States
  • Kaihua Hou
    Malone Center of Engineering in Healthcare, Johns Hopkins University, Maryland, United States
  • Patrick Herbert
    Malone Center of Engineering in Healthcare, Johns Hopkins University, Maryland, United States
  • Mathias Unberath
    Malone Center of Engineering in Healthcare, Johns Hopkins University, Maryland, United States
  • Jithin Yohannan
    Johns Hopkins Medicine Wilmer Eye Institute, Baltimore, Maryland, United States
  • Footnotes
    Commercial Relationships   Chris Bradley None; Kaihua Hou None; Patrick Herbert None; Mathias Unberath None; Jithin Yohannan None
  • Footnotes
    Support  5 K23 EY032204-02
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 2030 – A0471. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Chris Bradley, Kaihua Hou, Patrick Herbert, Mathias Unberath, Jithin Yohannan; Determining the Accuracy of Rates of RNFL Loss Measured by Peripapillary OCT. Invest. Ophthalmol. Vis. Sci. 2022;63(7):2030 – A0471.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose : To describe the rates of retinal nerve fiber layer (RNFL) loss in a large population of glaucoma patients and to understand how the accuracy of detecting certain rates of RNFL loss varies based on the frequency and temporal spacing on OCT exams.

Methods : Measurements of RNFL thickness using Zeiss OCT were obtained at a tertiary glaucoma practice from glaucoma and glaucoma suspect patients who were 18 years or older from April 1st, 2013 to July 30th, 2021. Only eyes with at least 5 measurements of superior and inferior RNFL thickness ≥ 50 m and signal strength (SS) ≥ 6 were included in the analysis.

Linear regression was performed on data from each eye and population level rates of RNFL change were computed. Simulated data were then generated from the regression slopes and residuals. The i th simulated data point from the k th eye had the form yi,k = βk + εi,k where εi,k was randomly chosen from the residuals of all eyes whose rates of worsening were within 0.25 μm/yr of βk. For each eye, M data points were generated within time period T under two conditions: in the "evenly spread" condition, simulated data were evenly spread out over time; in the "clustered" condition they were clustered (half and half) at the endpoints. The percentage of cases where the true slope was at or below the observed slope was calculated for different M and T.

Results : 12,150 eyes from 7,392 patients satisfied our inclusion criteria. The mean (SD) time between measurements was 403 (180) days. Figure 1 shows the rates of worsening, SS, and baseline RNFL thickness of our sample. The median, 75th and 90th percentile rates of worsening are –0.39, –1.09, and –2.35 μm/yr; the mean (SD) SS is 8.0 (1.0); and the mean (SD) baseline RNFL thickness is 87.50 (32.08) μm. Figure 2 shows the percentage of cases (y-axis) where an observed rate of worsening (x-axis) came from an eye whose actual rate of worsening was at least as severe as the observed rate. Clustering measurements and expanding the time interval between measurements improved accuracy, with the lowest accuracy in the [–3, –2] μm/yr range.

Conclusions : Our analysis shows that detecting rapid worsening (90th percentile) on OCT at 70% correct requires ~13 measurements clustered at the endpoints of a 2-year period. Current clinical practice is 1-2 OCT scans per year.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

 

×
×

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.

×