An Evaluation of Keratometry in 6-Year-Old Children.

The ability to determine corneal curvature or corneal refractive power with a high degree of accuracy and repeatability is important in both clinical and research situations. Reproducible keratometry is especially important for cataract surgery and in monitoring changes in corneal curvature over time. Many studies have documented changes or lack of changes in corneal power with age and its role in the development of myopia and progression of astigmatism. Findings in these studies need to be interpreted in light of the repeatability of the keratometers and other instruments used.

A variety of instruments capable of performing keratometry are in current clinical and research use. The IOLMaster (Carl Zeiss Meditec AG, Germany) and the RK-F1 AutoRef-Keratometer (Canon Inc., Japan) are two relatively new instruments. The IOLMaster semi-automatically performs measurements of axial length, corneal curvature, anterior chamber depth, and corneal diameter using non-contact techniques. The RK-F1 fully automatically performs keratometry and autorefraction.

Although a number of studies have reported excellent repeatability and accuracy of axial length measurement by the IOLMaster compared to ultrasound, few have assessed the repeatability of keratometry by this instrument, and no study has assessed the repeatability of keratometry when used in young children. To our knowledge, no reports have assessed the repeatability of keratometry by the RK-F1, or the comparability between the RK-F1 and the IOLMaster.

DISCUSSION

Repeatability

In the current study on 6-year-old children, both the IOLMaster and RK-F1 gave highly repeatable keratometry measurements, with similar limits of repeatability between the two instruments. There was greater variability of differences along the steepest meridian for both instruments, resulting in wider 95% limits of repeatability for the steepest meridian than the flattest meridian. This is probably a mathematical artefact arising from the use of the derived variable (refractive power), calculated from radius of curvature.

To our knowledge, the repeatability of keratometry by the RK-F1 has not been examined in previous studies, while those that examined keratometry by the IOLMaster did not examine its repeatability in children. Santodomingo-Rubido et al studied the repeatability of keratometry in a group of 52 adults with no ocular pathology using the vectorial method described by Thibos The mean differences in mean corneal radius and J0 and J45 vectors were all negligible. The 95% limits of repeatability for these variables ranged from 0.03 to 0.05 mm (equivalent to differences of approximately 0.17 to 0.29 D). However, direct comparison with our study is difficult due to the different methods of analysis.

Keratometry in the current study was performed before and after the instillation of cyclopentolate as there is little evidence to support an effect of cyclopentolate on its repeatability. In Heatley et al's study of patients at a cataract clinic, no change in mean keratometry along the flattest and the steepest meridians, or in mean keratometry was found on repeated measurements in the group that did not receive phenylephrine or tropicamide (34 subjects). In the group that did receive these mydriatics (47 subjects), a significant but small difference was found for mean keratometry along the steepest meridian (difference of group mean 0.08 D). Liang et al studied the effects of cycloplegia on repeatability by examining the change in the degree of agreement between the handheld Nikon Retinomax K-Plus and the on-table Topcon KR8100 autokeratometer in two cohorts of children (aged 2 to 12 years), one of which received cycloplegia. The authors found a significant change in agreement between the two instruments for the horizontal meridian of both eyes but only for the vertical meridian in the left eyes. However, the results of these studies may have been affected by the performance of applanation tonometry on the subjects and by interindividual variability that was not controlled for by the use of between-group comparisons. In contrast, Dobson et al used the Nikon Retinomax K-Plus keratometer to measure corneal curvature in 250 children aged 3-5 years and found no change in corneal curvature when measured with and without pupil dilation using cyclopentolate.

Comparability

The systematically steeper keratometry readings given by the IOLMaster compared to the RK-F1 represents approximately a one step change in refractive error correction. This difference was not related to the difference in estimates of principal axis. The 95% limits of agreement represent maximum deviations from the mean difference within which 95% of differences in measurements can be expected to lie. Thus, using the RK-F1 as the arbitrary reference, in 95% of measurements, the IOLMaster would give keratometry values along the flattest meridian that are as much as 0.08 D flatter or as much as 0.66 D steeper. Similarly, keratometry along the steepest meridian may be as much as 0.29 D flatter or 0.65 D steeper than the RK-F1.

To our knowledge, a comparison between the IOLMaster and RK-F1 has not been conducted before. However, the IOLMaster has been compared to the Javal-Schiotz keratometer and the EyeSys videokeratoscope. In that study, it was found that the IOLMaster gave steeper keratometry readings than did both the Javal-Schiotz keratometer (mean difference of mean corneal radius, -0.03 mm) and the EyeSys videokeratoscope (mean difference of mean corneal radius, -0.06 mm). These differences are equivalent to approximately 0.17 D and 0.34 D, respectively (calculated assuming one corneal radius is 7.70 mm and refractive index is 1.3375). In a study of 252 eyes of 134 patients aged 7 to 94 years, Nemeth et al also found that the IOLMaster gave significantly steeper keratometry readings than a Javal-type keratometer (mean difference ± SD of 0.17 ± 0.48 D). These differences are comparable to the differences with the RK-F1 found in the current study.

Systematically higher keratometry for both the flattest and the steepest meridians between the IOLMaster and RK-F1 may be due to the difference in the area of central cornea measured by each instrument. Previous studies have documented that the central cornea is more curved than the peripheral cornea. The RK-F1 measures a more peripheral part of the cornea than the IOLMaster because it uses a larger mire size.

Application

The findings of the current study have implications for clinical practice and research. Highly repeatable keratometry is desirable in both settings, especially when follow up is required. It should be noted that although there is no significant mean difference in keratometry on repeated measurements in both instruments, the 95% limits of repeatability are within 0.2 D (for the flattest meridian) or 0.35 D (for the steepest meridian) of the mean difference. This is important clinically and needs to be taken in consideration when repeated measures are performed over time.

The systematic difference of keratometry between the two instruments should be noted in situations when keratometry by these two instruments are being compared. In clinical practice, it should be noted that the IOLMaster may give keratometry values up to about 0.3 D flatter or 0.6 D steeper than the values given by RK-F1.

Conclusions:

Keratometry was highly repeatable for both the IOLMaster and RK-F1 instruments when used in young children. These instruments would be suitable for use in monitoring changes of corneal curvature over time. Small significant systematic differences in keratometry between the two instruments were also found.