The Asian Dataset; Normative Database Analyses - Zeiss CIRRUS HD-OCT 500 User Manual

The Asian Dataset

Normative Database Analyses

CIRRUS HD-OCT User Manual
In practice, clinicians and operators should quantitatively and qualitatively review scans
before comparing them to the CIRRUS RNFL or Macula Normative Databases. The
normative limits for scans that have poor scan quality should be used with caution.
The CIRRUS Asian RNFL and Macula Normative Databases were developed utilizing 315
subjects (aged 19–79). These normative databases have a similar gender distribution (159
males, 156 females). Ethnicity breakdown of the CIRRUS Asian RNFL and Asian Macula
Normative Databases is as follows: 44% Japanese, 44% Chinese, and 12% Indian.
Ganglion Cell Analysis normative data was developed using post–hoc analysis utilizing the
same analysis design described here. ONH normative data was developed using post–hoc
analysis with a different design, described in a later section.
Results revealed that the mean difference in the average RNFL thickness between any two
race groups is within 5 μm. Chinese have thicker mean average thickness, superior
quadrant average, and inferior quadrant average while Indians have the thickest
measurements for these parameters. The largest difference in the RNFL thickness between
two race groups is for the temporal quadrant average between Chinese and Indians, with a
difference of 15 μm.
Note that CIRRUS Asian RNFL, Asian Macula, and Asian Ganglion Cell Analysis Normative
Databases are adjusted only by age, not by ethnicity or any other parameter. The normative
limits provided for comparisons of individual data to the normative database do not take
into account differences that may be present due to ethnicity, axial length, refraction, optic
disc area, or signal strength.
From these scans the normative databases for RNFL, Macular Thickness and GCL + IPL
Thickness were created. The age range for all databases was from 19 to 79 years. Mean
age of the subjects was 47 years.
Regression model analyses were used to estimate the normative limit of each of the
CIRRUS HD-OCT RNFL, macular thickness, and ganglion cell analysis parameters adjusted
by age. For each fitted regression model, the residuals were derived for each eye by
subtracting estimated expected mean reading, ET(age0), from the measured or observed
reading, Obs(age0). In other words, residual = Obs(age0) – ET(age0). The goal was to
th
predict the 100x percentiles (NL, normative limit) of the residuals, so that the 100 x 1%
limit of the CIRRUS HD-OCT parameter readings could be estimated as follows:
ET (age 0) + NL(100 x 1%) < Obs (age0) (Equation A)
The 1st, 5th, 95th, and 99th percentiles of the residuals were estimated by the empirical
distribution of residual. Then the estimated 1%, 5%, 95% and 99% normal limits of
CIRRUS HD-OCT parameters for a normal subject with an age of age0 were established by
Equation (A). It should be noted that the study site effect was not considered in the
normative limits calculation since the objective was to establish the normative limits for the
general population.
Appendix
2660021169012 Rev. A 2017-12
A-27