## Purpose

To evaluate the repeatability, reproducibility, or both of the biometry and keratometry measurements obtained by a new optical low-coherence reflectometer and keratometer (Lenstar LS 900, version 1.10; Haag-Streit).

## Design

Prospective, comparative, observational study.

## Methods

__ setting: __Private practice, Lynwood, California, and Santa Monica, California.

__The measurements of the second eye to be operated on of 37 patients between October 2010 and January 2011 were analyzed.__

**study population:**__The axial length, central corneal thickness, aqueous depth, anterior chamber depth, crystalline lens thickness, white-to-white corneal diameter, as well as the keratometric readings at the flattest meridian, the steepest meridian, the average K, the amount of astigmatism, and the minus astigmatic cylinder axis were measured before surgery on the patient’s first eye and were repeated 1 month later before the second eye surgery.__

**observation procedure:**__The repeatability and reproducibility of the biometry and keratometry measurements.__

**main outcome measures:**## Results

Intrasession repeatability and intersession reproducibility were excellent with a very low coefficient of variation and high interclass correlation coefficients for all measured parameters. Bland-Altman plots show good correlation for axial length measurements (95% limits of agreement ranging from −0.056 to +0.04 mm), anterior chamber depth (−0.22 to +0.18 mm), crystalline lens thickness (−0.21 to +0.27 mm), corneal diameter (−0.28 to +0.24 mm), average keratometric readings (−0.56 to +0.47 diopters), and amount of astigmatism (−0.58 to +0.40 diopters).

## Conclusions

The precision of the measurements obtained by the new optical reflectometer and keratometer is extremely high.

The Lenstar LS 900 (Haag-Streit, Koeniz, Switzerland) is a noncontact optical reflectometer and keratometer that has been available for clinical use since 2009. It measures the axial dimensions of the eye in a single step. The technology is based on optical low-coherence reflectometry, with an 820-μm superluminescent diode. In addition to the axial length (AL), the unit allows the measurement of the central corneal thickness, aqueous depth, defined as the measurement from the corneal endothelium to the anterior lens surface; the crystalline lens thickness; and the white-to-white corneal diameter. A dual-zone automated keratometer analyzes the anterior corneal curvature at 32 reference points orientated in 2 circles at approximately the 2.3- and 1.65-mm optical zones. The keratometric (K) readings are calculated using a 1.3375 index of refraction and include the flattest meridian (K _{FLAT} ), the steepest meridian (K _{STEEP} ), and the minus astigmatic cylinder axis.

The purpose of this study was to evaluate the repeatability and reproducibility of the biometry and keratometry measurements obtained by the Lenstar LS 900 (version 1.10; Haag-Streit). Repeatability was defined as the intrasession variability obtained by using the same instrument and operator, and repeating the measurement during a short period. Reproducibility was defined as the intersession variability obtained by using the same measurement process among different instruments and operators, when the measurements are repeated over longer periods, or both. A low repeatability, low reproducibility, or both mean that the magnitude of the expected measurement is low, thus indicating high precision.

## Methods

This prospective, comparative, observational study analyzed the measurements of the second eye to be operated on of 37 patients seen between October 2010 and January 2011. Biometry included the AL, central corneal thickness, aqueous depth, lens thickness, and corneal diameter, and keratometry included K _{FLAT }, K _{STEEP }, and the astigmatic axis. The measurements were obtained initially through an undilated pupil before the surgery on the first eye (session 1) and then repeated by the same technician using the same unit approximately 1 month later before the surgery on the second eye (session 2). In addition, we calculated the following parameters needed for IOL power calculation: the anterior chamber depth (ACD), where ACD = central corneal thickness + aqueous depth; and the average K, where average K = (K _{FLAT }+ K _{STEEP })/2. We also calculated the magnitude of astigmatism (K _{STEEP }− K _{FLAT }), a measurement that has become of prime importance with the advent of premium intraocular lenses.

The Lenstar LS 900 is calibrated daily before obtaining the measurements. At each session, the patient is asked to fixate on a flashing red light, and as soon as the image of the eye on the computer monitor is in focus, a measurement is obtained. All measurements are repeated 5 consecutive times. The unit eliminates nonusable measurements and flags uncertain ones without eliminating them. The average value and a within-subject standard deviation (Sw) of each measurement are recorded. These repeated measurements were used to evaluate their intrasession repeatability at each session.

The measurements were repeated approximately 1 month later by the same technician, using the same unit. The technician had no access to the initial measurements when performing the second ones. The difference between the average measurements obtained at both sessions was calculated for each patient, along with the mean arithmetic difference (± standard deviation), the median absolute difference, the range of the differences, and the 95% confidence intervals (CIs) around the means. The difference in the average measurements between the 2 sessions was used to evaluate their intersession reproducibility.

## Statistical Analysis

Statistical analyses were performed using Microsoft Office Excel (Microsoft, Redmond, Washington, USA) and SPSS software for Windows, version 17.0 (SPSS Inc, Chicago, Illinois, USA).

Intrasession repeatability was assessed by analyzing the standard deviation of repeated measurements (Sw) and the coefficient of variation, which is defined as the ratio of the standard deviation of the repeated measurements to the mean. Lower Sw and coefficient of variation values indicate better repeatability.

Agreement between the measurements of the 2 sessions was evaluated by means of the method of Bland and Altman, who suggest plotting the differences between the measurements (y-axis) against their means (x-axis). The 95% limits of agreement (LoA) were defined as means ± 2 standard deviations of the differences between the measurements of the 2 sessions.

Evaluation of the intersession reproducibility between the data sets was accomplished by calculating the interclass correlation coefficients (ICCs). The ICCs are a measure of correlation or conformity for a data set when it has multiple groups. The ICCs range from 0 to 1, and a value of 0.9 or more indicates high agreement.

Measurements from the 2 sessions were compared using a paired *t *test because the data followed a Gaussian distribution according to the method of Kolmogorov and Smirnov. A *P *value of less than .05 was considered statistically significant. The median values are reported for the absolute differences, because these do not fit a Gaussian distribution.

## Results

Axial length for the 37 eyes ranged from 21.53 to 25.69 mm, with a mean value of 23.55 ± 1.03 mm, at session 1 and ranged from 21.60 to 25.70 mm, with a mean of 23.55 ± 1.04 mm, at session 2. The mean arithmetic difference between the 2 sessions was −0.008 ± 0.02 mm (95% CI, −0.015 to 0.00; *P *= .06). The absolute difference has a median value of 0.02 mm, with an ICC of 0.999. The Bland-Altman plot is shown in Figure 1 and reveals excellent agreement with a 95% LoA range between −0.056 and 0.04 mm.

Central corneal thickness ranged from 0.47 to 0.62 mm, with a mean value of 0.54 ± 0.04 mm, at session 1 and ranged from 0.46 to 0.62 mm, with a mean value of 0.54 ± 0.04 mm, at session 2. The mean arithmetic difference between the 2 sessions was 0.001 ± 0.006 mm (95% CI, −0.001 to 0.003; *P *= .53). The absolute difference has a median value of 0.004 mm with an ICC of 0.990.

Aqueous depth ranged from 1.99 to 3.34 mm, with a mean value of 2.61 ± 0.30 mm, at session 1 and ranged from 2.05 to 3.67 mm, with a mean value of 2.59 ± 0.32 mm, at session 2. The mean arithmetic difference between the 2 sessions was −0.024 ± 0.10 mm (95% CI, −0.055 to 0.007; *P *= .15). The absolute difference has a median value of 0.04 mm with an ICC of 0.949.

Anterior chamber depth ranged from 2.52 to 3.87 mm, with a mean value of 3.15 ± 0.29 mm, at session 1 and ranged from 2.59 to 3.67 mm, with a mean value of 3.13 ± 0.31 mm, at session 2. The mean arithmetic difference between the 2 sessions was −0.024 ± 0.10 mm (95% CI, −0.055 to 0.007; *P *= .16). The absolute difference has a median value of 0.04 mm with an ICC of 0.946. The Bland-Altman plot is shown in Figure 2 with a 95% LoA range between −0.22 and 0.18 mm.

Lens thickness ranged from 3.94 to 5.43 mm, with a mean value of 4.60 ± 0.43 mm, at session 1 and ranged from 3.94 to 5.42 mm, with a mean value of 4.63 ± 0.44 mm, at session 2. The mean arithmetic difference between the 2 sessions was −0.030 ± 0.12 mm (95% CI, −0.007 to 0.006; *P *= .14). The absolute difference has a median value of 0.04 mm with an ICC of 0.963. The Bland-Altman plot is shown in Figure 3 with a 95% LoA range between −0.21 and 0.27 mm.

The corneal diameter ranged from 11.08 to 13.09 mm, with a mean value of 12.03 ± 0.45 mm, at session 1 and ranged from 11.00 to 13.00 mm, with a mean value of 11.99 ± 0.43 mm, at session 2. The mean arithmetic difference between the 2 sessions was −0.047 ± 0.24 mm (95% CI, −0.121 to 0.028; *P *= .25). The absolute difference has a median value of 0.08 mm with an ICC of 0.849. The Bland-Altman plot is shown in Figure 4 with a 95% LoA range between −0.28 and 0.24 mm.

K _{FLAT }ranged from 40.80 to 48.40 diopters (D), with a mean value of 43.33 ± 1.76 D, at session 1 and ranged from 40.40 to 48.31 D, with a mean value of 43.34 ± 1.75 D, at session 2. The mean arithmetic difference between the 2 sessions was 0.005 ± 0.28 D (95% CI, −0.083 to 0.092; *P *= .92). The absolute difference has a median value of 0.19 D with an ICC of 0.987. K _{STEEP }ranged from 41.70 to 49.39 D, with a mean value of 44.29 ± 1.75 D, at session 1 and ranged from 41.79 to 49.58 D, with a mean value of 44.21 ± 1.77 D, at session 2. The mean arithmetic difference between the 2 sessions was −0.087 ± 0.29 D (95% CI, −0.176 to 0.003; *P *= .09). The absolute difference has a median value of 0.23 D with an ICC of 0.987. Average K readings ranged from 41.03 to 48.90 D, with a mean value of 43.81 ± 1.73 D, at session 1 and ranged from 41.09 to 48.89 D, with a mean value of 43.77 ± 1.74 D, at session 2. The mean arithmetic difference between the 2 sessions was −0.041 ± 0.26 D (95% CI, −0.121 to 0.039; *P *= .34). The absolute difference has a median value of 0.18 D with an ICC of 0.989. The Bland-Altman plot is shown in Figure 5 with a 95% LoA range between −0.56 and 0.47 D.

The amount of astigmatism ranged from 0.04 to 2.18 D, with a mean value of 0.961 ± 0.57 D, at session 1 and ranged from 0.05 to 2.16 D, with a mean value of 0.869 ± 0.56 D, at session 2. The mean arithmetic difference between the 2 sessions was −0.091 ± 0.25 D (95% CI, −0.188 to 0.004; *P *= .31). The absolute difference has a median value of 0.19 D with an ICC of 0.904. The Bland-Altman plot is shown in Figure 6 with a 95% LoA range between −0.58 and 0.40 D.