Posters

Benefit of optically measured crystalline lens data for IOL calculation of very short eyes

Poster Details

First Author: P.Hoffmann GERMANY

Co Author(s):    M. Abraham   T. Olsen   P. Preußner        

Abstract Details

Purpose:

To quantify the benefit of crystalline lens thickness data in IOL calculation of very short eyes.

Setting:

Private eye clinic in Germany

Methods:

The 100 shortest cataract and CLE eyes measured with our Haag-Streit Lenstar optical low coherence (OLCR) biometer between January 2011 and June 2013 were reviewed retrospectively. Alcon SA60AT was implanted 42x, SN60WF 13x, AMO Tecnis1 25x, others 20x. Preoperative data (axial length, corneal radii, corneal thickness, anterior chamber depth, lens thickness, refraction, IOL type and IOL power) were entered into three IOL calculation packages: Okulix 8.88 (ray-tracing), PhacoOptics 1.10.100.2017 (thick lens Gaussian + ray-tracing) and Holladay IOL consultant (Gaussian formula Holladay 2). Pupil width was assumed to be 3.0 mm. Postoperative refraction was predicted with and without lens thickness data. For comparison purposes, the old Holladay formula was also used. Prediction error was defined as achieved minus predicted spherical equivalent.

Results:

Mean axial length was 21.05 ± 0.62 mm, IOLs implanted were 26.5 - 38.0 dpt (Median 30.0). Mean ± standard deviation prediction error [D] with / without lens thickness (LT) was - 0.11 ± 0.73 / - 0.11 ± 0.74 for Holladay 2, - 0.14 ± 0.55 / - 0.29 ± 0.63 for Okulix and - 0.08 ± 0.51 / - 0.31 ± 0.61 for PhacoOptics. The number of refractive outliers > 1 D could be reduced from 16% to 6% for PhacoOptics, from 17% to 10% for Okulix but remained unchanged (21%) for Holladay 2. Means as well as variances differ highly significant (P < 0.01) between calculations with and without LT data for Okulix and PhacoOptics but not Holladay 2. Holladay 1 (+0.10 ± 0.72) delivers a result similar to Holladay 2 (means differ: t-Test P < 0.01, variance does not: F-test P = 0.78). The improvement achieved by LT data increases towards shorter axial length and higher IOL power. Mean absolute error improves from 0,58 to 0,57 for Holladay 2, 0,56 to 0,45 for Okulix and 0,56 to 0,42 for PhacoOptics when LT data is used.

Conclusions:

Both ray-tracing programs can significantly improve their predictive precision using crystalline lens thickness data. No relevant improvement can be seen in the Holladay 2 formula. Variances of the ray-tracing software packages are approx. 2 times smaller than those of Holladay 2 or Holladay 1 formulae when LT data is used. Therefore, refractive precision in short eyes can be vastly improved when Lenstar data sets are processed with Okulix or PhacoOptics. FINANCIAL INTEREST: One of more of the authors... gains financially from product or procedure presented