Sin Wan Cheung The Hong Kong Polytechnic University School of Optometry Hung Hom Hong Kong SAR NA CHINA E-mail: email@example.com
Background: Cycloplegia has been shown to have no effect on axial length measurement made with the IOLMaster in adults. The current study aimed at evaluating the effect of cycloplegia on axial length and anterior chamber depth (ACD) measurements made with the IOLMaster and an ultrasonic biometer in children.
Methods: Pre- and post-cycloplegic axial length and ACD were measured with the IOLMaster followed by the Sonomed A-5500 in 31 children aged from seven to 15 years by the same examiner. The 95% limits of agreement (LoA) were determined, if there were no significant correlations found between the mean differences and their means.
Results: Seven subjects were excluded. Results from the remaining 24 subjects show that the effects of cycloplegia, instruments, and interaction between cycloplegia and instrument on axial length measurement were insignificant (repeated measure ANOVA F1,23 < 2.19, p > 0.15). The 95% LoA in cycloplegia were better with the IOLMaster (-0.04 to 0.04 mm) than with the Sonomed A-5500 (-0.13 to 0.14 mm). The 95% LoA between the two instruments were similar with and without cycloplegia (pre-cycloplegia: -0.20 to 0.27 mm; post-cycloplegia: -0.17 to 0.22 mm).
There was no significant interaction between cycloplegia and instrument in ACD measurement (repeated measure ANOVA F1,23= 0.85, p = 0.37), however, ACD was 0.05 to 0.06 mm shorter before cycloplegia (repeated measure ANOVA F1,23= 44.70, p < 0.001) and was 0.06 to 0.08 shorter measured with the IOLMaster (repeated measure ANOVA F1,23= 28.81, p < 0.001).
Conclusion: Effects of cycloplegia on axial length measurement in children made with IOLMaster and Sonomed A-5500 were insignificant. In contrast, ACD measurement was significantly affected by cycloplegia and different instruments.
Ultrasonic biometry has been regarded as the gold standard in ocular biometry for many years for the measurement of corneal thickness, anterior chamber depth (ACD), crystalline lens thickness, vitreous chamber depth and the overall axial length.1 A-scan gives fairly accurate results with good reliability on ocular biometry. 1 The main drawback is the contact procedure, which requires the use of a topical anaesthetic and so is not easy to use. Advances in technology now allow ocular biometry measurement using optical pachymetry (for example, IOLMaster and Orbscan II), confocal microscopy (for example, Nidek ConfoScan), interferometry (example, anterior OCT) and the Scheimpflug principle (for example, Pentacam). These newer methods are superior to traditional ultrasonic biometry for several reasons. All except confocal microscopy allow non-contact procedures and the results are comparable or more repeatable than ultrasonic biometry.2–4 Among these new instruments, the IOLMaster (Zeiss, CA, USA) is the most studied and commonly used.
The IOLMaster is being used more frequently in routine examinations on a variety of patients in clinical practice, for example, for the determination of the intraocular lens power for cataract patients5–7 and for monitoring axial elongation in children.8,9 The major disadvantage is that unlike the ultrasonic biometry, the IOLMaster cannot be used for measuring corneal thickness, thickness of the crystalline lens and vitreous chamber depth.
The IOLMaster uses partial coherent interferometry in measuring axial length and optical pachymetry in measuring ACD.2–5,9–12 Both of these techniques may be affected by accommodation and ACD appears to be reduced during accommodation.13 In axial length measurement, the IOLMaster converts optical path length into geometrical length using an average refractive index. As the average refractive index changes with crystalline lens thickness, vitreous chamber depth and ACD during accommodation, using an average refractive index may overestimate axial length during accommodation.13,14
The IOLMaster produces repeatable and reliable results for axial length with4,9 and without2,3 cycloplegia. Measurements made with the IOLMaster are reported to be more repeatable than those made by ultrasonic biometry,2,4,9 however, most of these studies were performed on adults. As the axial length and ACD measurements with IOLMaster and ultrasonic biometry may be affected by accommodation, agreements between the non-cycloplegic and cycloplegic examinations and between the different instruments may also be affected.
The primary aim of this study was to evaluate the effect of cycloplegia on the measurements of axial length and ACD made with IOLMaster and ultrasonic biometry in children. A secondary aim was to determine the agreement between the two instruments with and without cycloplegia. Children were studied as axial length measurement is an essential examination in myopic control studies. Their accommodation is relatively more active than adults. Thus, the accuracy of axial length measurement in children may be affected by the use of cycloplegia.
Thirty-one children, aged between seven to 15 years, were recruited from the Optometry Clinic of The Hong Kong Polytechnic University. Children with accommodative or binocular problems, previous experience in contact lens wear or myopic control treatment or known allergy to eye-drops used in the current study were excluded. All subjects had normal general and ocular health. The purposes and procedures were fully explained to the subjects and their parents. Informed consent was obtained from the parents at the beginning of the study. The study followed the tenets of the Declaration of Helsinki revised in 2002 and ethics approval was obtained from the Departmental Research Committee of the School of Optometry of The Hong Kong Polytechnic University.
All examination was performed by one investigator. Preliminary eye examinations consisting of refraction, binocular vision, accommodation tests and assessment of the anterior and posterior ocular health were performed to exclude ineligible subjects. Ocular biometry using IOLMaster (Version 3, Zeiss Humphrey System, CA, USA) followed by ultrasonic biometry (Sonomed A-5500, Sonomed Inc, NY, USA) was performed with and without cycloplegia. Axial length and ACD were determined with the IOLMaster according to the manufacturer's instructions. Five consecutive readings of axial length were made for each subject. All readings had signal-to-noise ratio greater than 2.0 and the difference between the greatest and smallest readings was less than 0.20 mm. Three readings of corneal curvature and five readings of ACD were determined. Differences among the five ACD readings were all less than 0.15 mm.
Ocular biometric measurements were made with a Sonomed A-5500 biometer after instillation of a local anaesthetic (proparacaine hydrochloride 0.5%). The probe was sterilised with an alcohol swab and carefully applied to the surface of the cornea with minimal corneal pressure exerted. Subjects were asked to fixate a letter at six metres to control accommodation. Five consecutive readings were made with a standard deviation less than 0.1 mm.
The procedures were repeated after cycloplegia (cyclopentolate hydrochloride 1%). Measurements with IOLMaster were followed by the Sonomed A-5500 biometer when the amplitude of accommodation was less that 2.00 D (measured by the push-up test). Corneal integrity was re-examined at the end of the examination. No significant corneal staining was found on any subject before or after the examination.
Treatment of data
Data from the two eyes were highly correlated and so only the data from the right eyes were used for the analysis. Subjects were excluded if significant outliers (exceeding three standard deviations from the sample mean of all available data) were detected in either axial length or ACD measurement, between the two instruments and with and without cycloplegia. Distributions of the data obtained from the IOLMaster and Sonomed A-5500 biometer with and without cycloplegia were not significantly different from normal (Kolmogorov-Smirnov tests, p > 0.42). Therefore, parametric tests were used for data analysis. Repeated measures ANOVA was used to compare the axial length and ACD made by the IOLMaster and Sonomed A-5500 biometer with and without cycloplegia. If a significant difference was present, paired t-tests with Bonferroni correction were used to identify pairs that were significantly different, that is, an alpha level of 0.0125 instead of 0.05 was taken to be statistically significant. Pearson correlations were used to evaluate the relationship of the differences and their means measured by the two instruments with and without cycloplegia. The 95% limits of agreements (LoA) were determined using the Bland and Altman method.15–17
Data from seven subjects were excluded, two due to incomplete data and five due to significant outliers (exceeding three SD) in axial length and ACD measurements with and without cycloplegia. The demographic data of the remaining 24 subjects (eight male, 16 female) are shown in Table 1.
Table 1. Demographic data of the 24 children
Mean ± SD
10.5 ± 1.8
Refractive sphere (D)
-0.98 ± 2.84
Refractive cylinder (D)
-0.97 ± 1.01
Axial lengths measured by the IOLMaster and Sonomed A-5500 biometer with and without cycloplegia are shown in Table 2. Repeated measures ANOVA showed insignificant instrument effect (F1,23= 2.19, p = 0.15), insignificant cycloplegia effect (F1,23= 0.003, p = 0.96) and insignificant interaction effect between instrument and cycloplegia (F1,23= 0.004, p = 0.95). There were no correlations between the mean differences and their means measured by the two instruments with and without cycloplegia (Pearson 0.014 < r < 0.31, p > 0.14). The 95% LoA in axial length measurement with and without cycloplegia ranged from -0.04 to 0.04 mm for the IOLMaster and from -0.13 to 0.14 mm for the Sonomed A-5500 biometer (Figure 1). The 95% LoA between measurements made with IOLMaster and Sonomed A-5500 biometer were -0.20 to 0.27 mm without cycloplegia and -0.17 to 0.22 mm with cycloplegia (Figure 2).
Table 2. The mean ± standard deviation of axial length (mm) measured by the IOLMaster and the Sonomed A-5500 biometer and their mean differences, before and after cycloplegia
23.86 ± 1.20
23.82 ± 1.17
0.035 ± 0.12
23.86 ± 1.21
23.83 ± 1.18
0.030 ± 0.10
0.001 ± 0.02
0.006 ± 0.07
ACD values measured with the IOLMaster and Sonomed A-5500 biometer with and without cycloplegia are shown in Table 3. Repeated measures ANOVA showed significant instrument effect (F1,23= 28.81, p < 0.001) and significant cycloplegia effect (F1,23= 44.70, p < 0.001) but insignificant interaction effect between instrument and cycloplegia (F1,23= 0.85, p = 0.37). ACD was deeper when measured with the Sonomed A-5500 biometers and with cycloplegia (paired t-tests: p < 0.003). There were no correlations between the mean differences and their means measured by the two instruments with and without cycloplegia (Pearson -0.24 < r < -0.01, p > 0.27). The 95% LoA in ACD measurement with and without cycloplegia ranged from -0.02 to 0.11 mm for the IOLMaster and from -0.07 to 0.20 mm for the Sonomed A-5500 biometer (Figure 3). The 95% LoA between measurements made with IOLMaster and Sonomed A-5500 biometer ranged from -0.23 to 0.10 mm without cycloplegia and from -0.19 to 0.04 mm with cycloplegia (Figure 4).
Table 3. The mean and standard deviation of anterior chamber depth (mm) measured by the IOLMaster and the Sonomed A-5500 biometer and their mean differences before and after cycloplegia
3.59 ± 0.20
3.65 ± 0.21
0.06 ± 0.08
3.64 ± 0.19
3.71 ± 0.19
0.08 ± 0.06
0.05 ± 0.03
0.06 ± 0.07
Our results agree with a previous finding that the effect of cycloplegia on axial length measurement using IOLMaster is insignificant.4 The coefficient of repeatability of the mean differences between the pre- and post-cycloplegic axial length measured with the IOLMaster was 0.02 mm, which is comparable to the repeatability coefficients found in previous studies in children and adults.2,4,9 Hence, cycloplegia is not necessary for axial length measurement in normal adults and children using the IOLMaster. However, the result should not be generalised to cataract or aphakic patients, where the presence of lens opacities or the absence of a crystalline lens will affect the accuracy of IOLMaster.13,14
Theoretically, axial length measurement using the IOLMaster is affected by cycloplegia. Drexler and co-workers14 found that accommodation of four to five dioptres could increase axial length by 0.013 and 0.005 mm in emmetropic and myopic patients, respectively, using partial coherence interferometry. Using the Gullstrand's No. 1 schematic eye, Atchison and Smith13 calculated that accommodation of 10.9 D could induce a measurement error of 0.018 to 0.026 mm in axial length, which is equivalent to less than 0.10 D change in refractive error. This potential measurement error induced by accommodation was too small to affect axial length readings in adults and children with clear crystalline lenses. The result of our study is in agreement with their study that cycloplegia has minimal effect on axial length measurement.
Cycloplegia affects ACD as the crystalline lens displaces into the anterior chamber during accommodation, hence a longer ACD is expected after cycloplegia. Our results showed that after cycloplegia, ACD was deeper by 0.05 mm. Similar results have been reported in adults using IOLMaster.4 The IOLMaster uses optical pachymetry, while the Sonomed A-5500 uses ultrasonography in measuring ACD. Possible sources of error in ACD measurement with these instruments include the different reference planes of measurement, proximal accommodation and off-axis measurement induced in optical pachymetry and excessive indentation of the cornea in ultrasonic biometry. In this study, the influence of these factors on cycloplegic ACD measurement was controlled by having the same experienced examiner performing the tests with and without cycloplegia.
Agreements in axial length and ACD measurements between IOLMaster and ultrasonic biometry or other instruments incorporating different techniques (for example, Pentacam and Orbscan II) have been extensively studied. In general, the axial length measurement of IOLMaster is comparable with other techniques but with better repeatability and reproducibility.2–5,9,11,18 Agreement in ACD measurements between instruments is poor, as substantial systematic variations have been reported.
We found that the IOLMaster gave shallower ACD readings by 0.06 to 0.08 mm when compared to the Sonomed A-5500, both with and without cycloplegia. A similar observation was noted in our previous study using the same instruments.2 In contrast, using the same IOLMaster but a different ultrasonic biometer, another study group from the same institution reported deeper ACD measured with the IOLMaster.3 Thus, different ultrasonic biometers may give different ACD results,19 that is, while some ultrasonic biometers may give longer ACD measurement than the IOLMaster, others may give shorter ACD. Practitioners should be aware of the systematic difference between different models of ultrasonic biometers and its possible effect on the agreement with the IOLMaster.
Although both IOLMaster and ultrasonic biometry have reasonably good repeatability in ACD measurements, the repeatability is not as good as in axial length measurement. Instruments that may give better repeatability for ACD measurements are the AC-Master (Carl Zeiss Meditec AG, Germany) and the anterior segment OCT (for example, Visante, Zeiss, Germany).11,18
In conclusion, our results showed that axial length measurements made with the IOLMaster and Sonomed A-5500 ultrasonic biometer were not affected by the use of cycloplegia in children. Good agreement in axial length measurements between the two instruments with and without the use of cycloplegia was observed. ACD measurements with either instrument were shallower by 0.05 to 0.06 mm without the use of cycloplegia. There was a systematic difference in ACD measurement between the IOLMaster and the Sonomed A-5500 biometer, the former giving shorter measurement by 0.06 to 0.08 mm.
GRANTS AND FINANCIAL ASSISTANCE
This study was supported by a grant from The Hong Kong Polytechnic University.