Zeiss 3000 - Age Coefficient

Stratus OCT User Manual PN 2660021134133 A
Normative Database for Macula and RNFL Measurements
C-3
variation in the distribution of thicknesses does not change with the age of the population.
The distribution was then transformed into a Gaussian distribution, which was then used to
calculate the 1%, 5%, 95% and 99% limits of the normal population. A detailed statistical
description of the data transformation follows:
A standard simple linear regression analysis was used in the data analyses. For each of the
128 data points of the six scan lines, the expected thickness (ET) was estimated as a linear
function of age, i.e., Expected Thickness = a + b X age, where a and b were constants. The
values of a and b could differ among the regression lines for the 128 data points of the six
scan lines. The Normal Limit was then derived based on the percentiles of the residuals,
i.e., the difference between the estimated expected thickness (ET) and the observed
thickness (rt), as follows:
ET(age) + NL(5%) < rt(age) < ET(age) + NL(95%),
Where
NL(5%) and NL(95%) were the estimated 5th and 95th percentiles of the residuals,
and

ET(age) + NL(5%) and ET(age) + NL(95%) were the 5% and 95% Normal Limits.
In order to better estimate the percentiles of the residuals, Manly’s exponential
transformation
1
was used to transform the residuals so that the transformed residuals were
normally distributed. The percentiles of the transformed residuals were estimated based on
the normal distribution. Then the estimated percentiles of the residuals were derived by
converting the percentiles of the transformed residuals back to the original scale.
Age Coefficient
Analysis of subject demographics determined that expected thickness was dependent upon
age, but not significantly dependent upon other variables, i.e., right vs. left eye and
gender. Thus age correction is incorporated into the calculated results. Subject ethnicity
was self-reported by the subjects in the population comprising the normative database but
was not used as a variable in constructing the macular normative database.
1. Manly B. F. J. (1976) Exponential Data Transformations. The Statistician. Vol. 25, No. 1, pp
37-42.