Transcript PREDICTING AXIAL LENGTH USING AGE, KERATOMETRY AND

PREDICTING AXIAL LENGTH USING
AGE, KERATOMETRY AND
CYCLOREFRACTION
LIONEL KOWAL & PRASHANT SAHARE
Center for Eye Research Australia & Ocular Motility
Clinic, Royal Victorian Eye and Ear Hospital,
Melbourne, Australia
ABSTRACT
We measured axial length in 51 consecutive patients
having strabismus surgery.
We found significant correlation between axial length
and the three factors of age, keratometry and
cyclorefraction. We derived a formula which predicts
axial length with an accuracy of ≤±1.0 mm in 88% of
patients
Axial length = 34.643 + 0.0513 age – 0.2834
keratometry [DS] – 0.3663 cyclorefraction.
INTRODUCTION
Strabismus surgeons derive surgical doses from tables
describing millimetre doses of surgery. A given mm
repositioning of a muscle can be expected to have a
larger angular effect on a small globe and a smaller
effect on a large globe, yet surgical dose tables do not
take globe size into account.
Several attempts to modify surgical doses by considering
globe size have had inconclusive outcomes.
One difficulty of research in the area is obtaining reliable
axial length measurements in the relevant age group.
In our study we asked: Can we predict axial length
using age, keratometry and cyclorefraction?
MATERIALS AND METHODS
Charts were reviewed of all strabismus
patients on whom axial length measurements
were made in the previous 6 months, either inoffice with the Zeiss IOL Master® or by
ultrasound under anaesthesia. Fifty one
consecutive charts with complete data were
included in this study.
Keratometry was measured with the Nidek
KM-500® autokeratometer; averaged
keratometry was used for analysis. Refractive
error was measured using streak retinoscopy
after cycloplegia with cyclopentolate; spherical
equivalent readings were used for analysis.
The data was analysed by linear regression analysis.
Diagnostic plots including the normal quantile - quantile plot
and residual vs. predicted value plot were analysed.
RESULTS
Fifty one eyes were measured.
Age range: 6 months to 55 years (mean 9.9 y).
Axial lengths: 18.22 - 26.36 mm (mean 22.26)
Keratometry: 38.25 - 47.5D (mean 43.04D).
Cyclorefraction: –7.25 to +9 DS (mean +1.8DS).
The estimated coefficients for age,
keratometry and cyclorefraction were all
statistically significant (p <0.001).
RESULTS
Correlation between Axial Length, Age, and Refractive Error
Patient group
All patients
All patients
Age<2 years
Age>2 years
Age<10 years
Age >10 years
All myopes
All hyperopes
factors correlated
correlation coefficient
axial length and age
axial length and refractive error
age and axial length
age and axial length
age and axial length
age and axial length
axial length and refractive error
axial length and refractive error
0.7956
-0.8168
0.8868
0.7843
0.2550
0.7419
-0.2696
-0.6637
p value
0.0001
0.0001
0.0449
0.0001
0.1519
0.0004
0.6054
0.0001
Fig 1
In a linear regression analysis, the data fitted the model well and explained 87% of
variation (r2=0.87).The equation for predicting axial length for a given age, keratometry
and cyclorefraction is
Axial length = 34.643 + 0.0513 age – 0.2834 K – 0.3663 C
•
•
Quantile- quantile plot
Fig 2
The distribution of axial length is normally distributed as shown in probability and
quantile- quantile plot. The diagnostic plots, including the normal quantile-quantile
plot below and the residual vs. predicted value plot (Fig 3) did not show any apparent
departure from the assumptions of the linear model.
Fig 3 Residual vs. predicted value
In 53% of patients [27/51] the axial length predictions were within 0.5mm of
the measurement, in 35% [18/51] within 0.5 – 1.0 mm of the measurement and
in 12% [6/51] within 1.0-1.5mm.
•
•
DISCUSSION
In correcting an angular misalignment with strabismus surgery, we change
muscle position / length linearly. Various authors have suggested that we
consider globe size when we choose a surgical dose. Such measurements
are difficult in younger children and may require general anaesthesia.
In our series, regression analysis derived a formula [based on age,
keratometry, and cyclorefraction] suitable for estimating axial length.
In a remarkably similar series of patients, Kushner had previously shown
that examining age and refractive error and assuming an average
keratometry of 43.5D for all patients produced greater variation in axial
length predictions than we found.
He found 41% of predicted lengths to be within 0.5mm of measured
lengths [we found 53%] and 79% within 1mm [we found 88%].
The multiple regression analyses in his series resulted in r2 values ranging
from 0.38 to 0.49 c.f. 0.87 in ours. It seems that accurate measurement of
keratometry is the necessary extra factor for reliably predicting axial length.
CONCLUSION
For those who wish to modify
strabismus surgical dose measurements in
larger and smaller globes, axial length can
be predicted with considerable accuracy
using a formula involving age, keratometry
and cyclorefraction
REFERENCES
• Gillies and McIndoe
The use of ultrasonography in determining the amount of extra ocular
muscle surgery in strabismus
Australian Jnl of Ophthalmology 1982;10:191-194
• Kushner, Luchese and Morton
Variation in Axial Length and Anatomical landmarks in Strabismic
Patients
Ophthalmology 1991: 98: 400-406
• Kushner, Luchese and Morton
The influence of axial length on the response to strabismus surgery
Arch Ophthalmol 1989;107:1616-1618
• Kushner, Qui, Lucchese and Fisher
Axial length estimation in strabismic patients
JPOS 1996;33:257-261