Precision of a new ocular biometer in children and comparison with IOLMaster

To assess the repeatability and reproducibility of AL-Scan in agreement with those by the IOLMaster in healthy children, two skilled operators measured ocular parameters in 58 children. The parameters included keratometry (K) values, anterior chamber depth (ACD), axial length (AL), central corneal thickness (CCT), pupil diameter (PD), and corneal diameter (CD). The cohort comprised of 32 boys and 26 girls. The AL-Scan measurements showed high repeatability, as the test-retest repeatability (TRT) values of AL, CCT, ACD, Kf, Ks, Km, CD, and PD were 0.09 mm, 5.1 μm, 0.04 mm, 0.28 D, 0.24 D, 0.21 D, 0.39 mm, and 0.22 mm, respectively. The within-subject coefficient of variation (CoV) was low (<0.35%) and the intraclass correlation coefficients (ICC) of all parameters were >0.85. The interobserver reproducibility was excellent with low values of TRT and ICC > 0.95. The CoV of AL, CCT, ACD, and K was <0.22%. The 95% limits of agreement between the AL-Scan and the IOLMaster were narrow for all parameters except for CD. The repeatability and reproducibility of the new biometer, Al-Scan, was excellent for all parameters and can be routinely used in children to measure the biometric values.

Interobserver Reproducibility. Table 2 shows interobserver reproducibility outcomes for biometric measurements obtained using the AL-Scan. All parameters revealed high reproducibility with low S w values. The CoV of AL, CCT, ACD, and K values were lower than 0.22%; the highest CoV (6.68%) was observed for the astigmatism magnitude. The ICC values were>0.95 for all measurements. Table 3 shows the differences between the AL-Scan and IOLMaster. Statistically significant differences were observed between the AL-Scan and IOLMaster with respect to all the parameters except AL, J 0 , and J 45 . Although these differences were statistically significant, the  Table 2. Interobserver reproducibility outcomes for biometric measurements obtained using AL-Scan partial coherence interferometry in children. AL = axial length, CCT = central corneal thickness, ACD = anterior chamber depth, Kf = flattest keratometry, Ks = steepest keratometry, Km = mean keratometry, CD = corneal diameter, SD = standard deviation, Sw = within-subject standard deviation, TRT = test-retest repeatability (2.77 Sw), CoV = within-subject coefficient of variation, ICC = intraclass correlation coefficient. 95% LoAs were narrow for all parameters except for CD, whose 95% LoAs were between −1.10 mm and 0.26 mm (Figs 1-6).

Agreement Between the AL-Scan and IOLMaster.
Intraocular Lens Power Calculation. The calculated mean IOL power was 19.64 ± 3.20 D with the AL-Scan and 19.64 ± 3.19 D with the IOLMaster. The mean difference between the two biometers was 0.13 ± 0.14 D. The 95% LoAs were narrow (−0.15 to 0.41 D). The largest differences between the two biometers were 0.44 D.

Discussion
AL is a major factor for the calculation of intraocular lens (IOL) power in pediatric cataract surgery. An AL difference of 1 mm leads to a refractive error of approximately 2.7-3.0 D 16,17 . In addition, estimating the progression of myopia also depends on accurate AL measurement. Sahin et al. 18 15 found that the TRT of AL, as measured by the AL-Scan, was 0.05 mm in adults, thereby indicating excellent intraobserver repeatability. In the current study, the TRT of AL was low (0.09 mm); therefore, we speculate that the AL measured by the AL-Scan in children had excellent repeatability, similar to that in adults. In addition, the TRT of AL was 0.06 mm in reproducibility outcomes, which showed high interobserver reproducibility in a pediatric population. Agreement between the AL-Scan and IOLMaster was high for the AL measurement with a mean difference of 0.00 ± 0.03 mm, and the 95% LoAs were between −0.05 and 0.05 mm. The mean difference was smaller than that observed in the study between the Lenstar and IOLMaster by Hoffer et al. 19 , who reported a mean difference of 0.03 mm. Jasvinder et al. 20 also indicated that the results of Lenstar and IOLMaster were consistent in cataract patients (95% LoA was −0.04-0.07 mm) with respect to the current study. A few previous studies on cataract patients displayed a robust agreement between the AL-Scan and IOLMaster 10,15,21 . Therefore, we can consider that the AL-Scan and IOLMaster provided interchangeable AL measurements in a pediatric population. However, one of the major obstacles for ocular measurements in children is fixation. If the child does not stare at the target well,  Table 3. The mean difference, paired T-test, and 95% limits of agreement (LoA) for differences between the AL-Scan and the IOLMaster partial coherence interferometry in children. AL = axial length, ACD = anterior chamber depth, Kf = flattest keratometry, Ks = steepest keratometry, Km = mean keratometry, CD = corneal diameter, SD = standard deviation. unreliable AL measurements may be obtained. The AL-Scan can provide a series of ocular parameters by a single measurement, thereby facilitating the cooperation of the children for reliable results. Regarding ACD, a statistically significant difference with respect to the IOLMaster was found, with a relative higher mean value measured by the AL-Scan. However, a difference of 0.17 mm was not clinically relevant. This difference might be attributed to the usage of a lateral slit by IOLMaster to measure the ACD, whereas the AL-Scan is based on the Scheimpflug principle. Also, the IOLMaster and its previous versions have been reported to measure the lower ACD values than the other optical biometers in previous studies 10, [22][23][24] .
The AL-Scan measures CCT using the Scheimpflug principle, whereas the IOLMaster does not acquire this parameter. The high ICC (0.997 or 0.999) and the low TRT showed that this instrument can provide highly repeatable and reproducible measurements of CCT. Moreover, the ICC values were higher than those in the study by Kola et al. and the TRT values were less than that in our previous study 9,15 . The study by Chen et al. 25 , Huang et al. 26 , and Savini et al. 27 discovered that the ICC for the rotating Scheimpflug imaging was similar to the current study.
With low TRT (<0.28 D for repeatability and 0.21 D for reproducibility) and low CoV (<0.23% for repeatability and 0.17% for reproducibility), we found that AL-Scan provided highly repeatable and reproducible results for K values. The TRT of K values and astigmatism magnitude were <0.5 D, which was in the clinical range. In addition, similar results had been reported by Srivannaboon et al. 28 in cataract patients. The 95% LoA showed high agreement in Kf and Ks between the AL-Scan and IOLMaster, although the mean Ks value was higher with the IOLMaster than that with the AL-Scan, which might induce slight selection bias in the calculation of IOL power, thereby necessitating constant optimization. Nevertheless, the agreement of corneal power and astigmatism was clinically acceptable. The reason for the difference could be attributed to the usage of a double-ring by AL-Scan to measure keratometry, which could provide more data than the IOLMaster. Moreover, the IOLMaster provided slightly steeper K values than the AL-Scan. Similar findings have been reported when the IOLMaster was compared to other devices such as Lenstar 29 .  Repeatability and reproducibility of CD and PD were high in children, thereby confirming the results previously reported with the same biometer by Kola et al. in adult patients 9 . The study revealed that the agreement for CD was not satisfactory, since the 95% LoA was between −1.10 and 0.26 mm. Both devices used the same principle to assess CD; however, the AL-Scan took advantage of the 525-nm LED measuring the CD that could avoid the blurred images caused by wide-spectrum laser. Although the consistency of the CD was not as satisfactory as other parameters, the 95% LoA values were narrower than those in the previous studies 15,21 . This might be due to the clearer corneal limbus in children than the elders suffering from cataract.
The present study has some limitations. First, we did not include the patients suffering from pathological myopia, which might influence the cooperation. Second, individuals with keratoconus eyes were not enrolled. These drawbacks necessitate future investigations considering these parameters.
In conclusion, we found that the new optical biometer presented excellent intraobserver repeatability and interobserver reproducibility for all parameters in children. Therefore, the AL-Scan can be routinely used in children to measure the biometric values. Moreover, the AL-Scan and IOLMaster were highly consistent with respect to a majority of the parameters.

Subjects. This prospective study consisted of normal children at the Eye Hospital of Wenzhou Medical
University, Wenzhou, China. Children with ophthalmic surgery, dry eye, corneal disease, contact lens wear, strabismus, amblyopia, or ocular trauma were excluded. All eyes went through comprehensive ophthalmologic examinations before measurements, including subjective refraction, noncontact intraocular pressure measurement, slit-lamp microscopy examination, and ophthalmoscopy. The research was approved by the Board of Eye Hospital of Wenzhou Medical University, which adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from the childrens' parents after explaining the purpose of the research.  Instruments. The AL-Scan (software V.1.03) is based on PCI, measures the AL through an 830-nm infrared laser diode 12 . The optical biometer takes advantage of the light-emitting diode (LED) to assess the K values, CD, and PD 8 . The K values were measured at the 2.4 and 3.3 mm diameters 10 . In the current study, K values at the 2.4 mm diameters were recorded, as it is similar to that used by the IOLMaster. Then, the K values were calculated from the anterior corneal radius using the keratometric index of 1.3375, which included the flattest keratometry (Kf), the steepest keratometry (Ks), and the mean keratometry (Km). The measurement of CCT and ACD uses the Scheimpflug principle 9 . During a measurement, the AL-Scan performs in an automated mode measuring all the biometric variables.
The IOLMaster (version 5.4) is based on dual-beam PCI principle and utilizes a 780-nm infrared laser diode to measure the AL 10 . The ACD, defined as the length between the corneal epithelium and the anterior surface of the lens, is measured by lateral slit illumination 15 . The K values were calculated by analyzing the data from a hexagonal array of 6 points reflected off the surface of the cornea in the optical zone of an approximately 2.5 mm diameter 10,30,31 . CD was assessed by 3 consecutive measurements. Measurement Technique. Children were measured in a random order in order to avoid the methodological bias. 1% cyclopentolate hydrochloride eye drops were used (1 drop every 5 minutes, 3 times). After 40-60 min, the disappearance of the light reflex was confirmed in both eyes. For the study of repeatability, the children's right eyes were examined by the same examiner, who took 3 consecutive measurements. Subsequently, two skilled operators used the AL-Scan randomly in order to evaluate the interobserver reproducibility. Agreement between the AL-Scan and IOLMaster was assessed by the same observer who tool 3 measurements consecutively with every optical biometer. The measurements were conducted between 10:00 a.m. and 5:00 p.m. minimize the diurnal variations of corneal shape and thickness 32,33 . According to the operation manual, each subject was instructed to sit in front of the instrument, place the chin on the chinrest, fix the target, and adjust the position to mark the outer canthus of the patient and the horizontal lines of IOLMaster at the same height. Then, the children were asked to blink before examination to minimize the effect of tear film irregularities. The measurement for each child was completed within 30 min. Only high quality and eligible measurements were selected for further analysis and unreliable data were excluded. Intraocular Lens Power Calculation. In order to assess the effect of measurements difference in the clinical setting, we calculated the intraocular lens (IOL) power using the K and AL values from both devices. The SRK/T formula, with an A-constant of 119.0, was chosen for this purpose 34 .
Statistical Analysis. All data were analyzed using SPSS for Windows software (version 21, IBM Co., USA) and MedCalc statistical software (version 13.0, MedCalc Software Inc., Belgium). A P-value < 0.05 was considered statistically significant. Normal distribution was assessed by the Kolmogorov-Smirnov test, and a P < 0.05 was considered normally distributed. The results of all parameters were presented as the mean ± standard deviation (SD). The intraobserver repeatability and interobserver reproducibility of the AL-Scan were analyzed using the within-subject SD (S w ), test-retest repeatability (TRT), within-subject coefficient of variation (CoV), and intraclass correlation coefficients (ICC). The TRT = 2.77S w denoted the interval within which, the 95% of the differences were located. The CoV was calculated as the S w divided by the mean of all measurements; low values revealed better precision. In addition, if the ICCs were closer to 1, the reliability was higher. The paired t-test and Bland-Altman plots with 95% limits of agreement (LoA) were used to evaluate the agreement between the AL-scan and IOLMaster. The 95% LoA was calculated as the average difference between the AL-Scan and IOLMaster ± 1.96SD 35 . The narrower the 95% LoA, the better the agreement.