To determine the interobserver and intraobserver reproducibility of a Fourier-domain optical coherence tomography device (Cirrus HD OCT; Carl Zeiss Meditec, Dublin, California, USA) in normal pediatric eyes.
Prospective cross-sectional study.
One hundred healthy children were recruited prospectively and consecutively. Only 1 randomly chosen eye per subject was included in the study. The eye underwent 3 scans centered on the optic disc and another 3 scans centered on the macula that were acquired by a single operator. A fourth examination was performed by a second operator. Interobserver and intraobserver reproducibility were described by intraclass correlation coefficients (ICCs) and coefficients of variation (COVs).
The mean age was 9.15 years (range, 6.22 to 11.31 years; standard deviation, 1.05 years). Mean retinal nerve fiber layer thickness was 99.53 μm (standard deviation, 10.10 μm), and mean macular thickness was 282.91 μm (standard deviation, 11.83 μm). All the parameters evaluated were highly reproducible. Intraobserver COVs of the retinal nerve fiber layer measurements ranged from 2.24% to 5.52%, and the COV of macular thickness was 0.97%. The intraclass correlation coefficient was greater than 0.8 for all the parameters. The interobserver COV ranged from 2.23% to 5.18%, and the COV of macular thickness was 0.82%. In all the evaluated parameters, the intraclass correlation coefficient was more than 0.75. Repeatability was slightly better in children older than 10 years than in children younger than 9 years.
Retinal nerve fiber layer and macular measurements obtained by Fourier-domain optical coherence tomography showed good repeatability for healthy eyes in the pediatric population. Cirrus HD OCT examinations of the retina are reliable in children.
In recent years, several devices that allow an objective and quantitative evaluation of retinal structures have become useful in clinical practice because of the need for objective tests that complement funduscopy or photographs, especially in children, in whom they may prove difficult to perform. One of instrument that is becoming increasingly popular in pediatric ophthalmology is optical coherence tomography (OCT).
OCT is a noninvasive, noncontact method that uses low-coherence interferometry to perform high-resolution cross-sectional imaging of tissue morphologic features, providing an optical biopsy. The latest OCT devices, Fourier-domain (FD) OCTs, offer increased resolution compared with time-domain instruments. Current commercially available OCT instruments provide improved axial image resolution, between 5 and 7 μm, compared with earlier generations of OCT that ranged between 8 and 10 μm. A number of publications have proved the feasibility and validity of this technique in different clinical applications in children for diseases such as glaucoma, retinopathy of prematurity, and neurofibromatosis type 1.
All the OCT devices have an integrated normative database, which includes only individuals 18 years of age and older. To evaluate changes in retinal measurements accurately, it is first necessary to determine the range in the normal population and to quantify the accuracy, reproducibility, and repeatability of measurements made by the system. The objectivity and reproducibility of Cirrus OCT have been proven by several authors in adults, but never in children. The purpose of the present study was to examine the interobserver and intraobserver reproducibility of repeated measurements of the retinal nerve fiber layer (RNFL), optic disc, and macular measurements with an FD OCT in healthy children.
This study was undertaken in an elementary school from December 2010 through March 2011 as part of the Environmental Fetal Factors in the Development of the Optic Nerve and the ReTina study (EFFORT). From the 598 eligible children, 358 were included in the study, giving an acceptance rate of 60%. One hundred healthy children were recruited prospectively and consecutively from among the 358 for the reproducibility and repeatability study.
All subjects underwent a comprehensive ophthalmologic evaluation that included monocular visual acuity (HOTV chart read at 300 cm), stereopsis assessment (TNO test at 40 cm), ocular motility evaluation, and retinal and optic nerve assessments by means of OCT (Humphrey Zeiss Instruments, Dublin, California, USA). All children with abnormal results (visual acuity worse than 20/25, strabismus, or history of ocular diseases) were examined under cycloplegic mydriasis and the refractive error was measured.
The RNFL and optic disc measurements were obtained using the optic disc cube 200 × 200 protocol. Under this protocol, a 3-dimensional cube of data is generated over a 6-mm 2 grid of 200 horizontal scan lines, each composed of 200 A-scans. A Cirrus software algorithm automatically detects the center of the optic disc from this volume scan and positions a 3.46-mm diameter calculation circle over this point. From the 256 A-scans along this circle, the borders of the RNFL are delineated and thickness is calculated at each point along the circle. The macular images were obtained using the macula cube 200 × 200 protocol. This protocol generates 200 × 200 volume cube images with 200 linear scans performed by A scans and analyzes a cube 6 mm in diameter around the macula. Retinal thickness was calculated using the built-in macular analysis software on the Cirrus device, which is determined automatically by taking the difference between the inner limiting membrane and retinal pigment epithelium boundaries.
The examinations were performed without pupil dilation. Only scans with 7 strength/signal or more were accepted. Images with artifacts or missing parts were excluded and repeated. Internal fixation was used to suppress ocular movements, because it results in the highest reproducibility. Three scans were performed by a single operator (I.A.), and a fourth scan was performed by a second operator (N.E.). Between scans, subject position and focus were disrupted randomly, and alignment parameters had to be newly adjusted at the start of each image acquisition. The parameters collected were mean RNFL thickness (360 degrees), RNFL thickness in 4 quadrants (inferior, superior, nasal, temporal), ring area, disc area, cup-to-disc area ratio, macular thickness, and macular volume.
To be enrolled in the project, written informed consent was given by the parents or guardians of the child. The only exclusion criterion was the absence of a signed informed consent. Because we wanted to examine a sample representative of the normal pediatric population, no other exclusion criteria were considered. Statistical analyses were carried out with the Statistical Package for the Social Sciences version 15.0 (SPSS, Inc, Chicago, Illinois, USA).
Interobserver and intraobserver reproducibility were described by intraclass correlation coefficients (ICCs) and coefficients of variation (COVs). For the intraobserver study, we compared the results from the first 3 examinations performed by the same examiner (I.A.). For the interobserver study, the first and the fourth examinations were compared because they were performed by 2 different examiners (I.A., N.E.).
COVs were calculated as the standard deviation divided by the average of the measurement value, expressed as a percentage. A COV of less than 10% was considered reproducible and a COV of less than 6% was considered highly reproducible. The ICCs were considered to provide slight reliability (values between 0 and 0.2), fair reliability (values between 0.21 to 0.4), moderate reliability (values between 0.41 and 0.6), substantial reliability (values between 0.61 to 0.8), or almost perfect reliability (values of more than 0.81).
COVs obtained in different groups of children were compared according to their age. Children were divided into categories younger than 9 years (34 children), from 9 to 10 years (36 children), and older than 10 years (30 children). COVs were compared by analysis of variance with the Bonferroni post hoc test. P values less than .05 ( P < .05) were considered statistically significant. For multiple comparisons, we used the Bonferroni post hoc correction.
One hundred children were included in the study (52 boys and 48 girls). The age of the patients ranged from 6.22 to 11.31 years, with a mean age of 9.15 years. Mean best-corrected visual acuity (logarithm of the minimal angle of resolution) was −0.01 (20/20 Snellen equivalent), with a range from 0.3 to −0.2. The refractive errors ranged from −3.00 to +4.50 of spherical equivalent. Stereoacuity was full (60 seconds of arc or better) in 94 children and was reduced (worse than 60 seconds of arc) in 6 children. Twenty-nine of 100 children (29%) demonstrated ocular motility anomalies (21 exophoria (21%), 6 endophoria (6%), 2 esotropia (2%)). No patient from the study had to be excluded because of bad OCT image quality.
Table 1 shows the results obtained in the global and sectorial analysis of the RNFL and macula, measured by OCT by observer 1 in the first, second, and third scans and by observer 2 in the fourth scan. None of the differences between measurements may be considered as statistically significant after Bonferroni correction for multiple comparisons. As expected, quadrant distribution followed a double-hump pattern, being higher for the inferior quadrant, followed by superior, nasal, and temporal quadrants. Mean differences between the measurements obtained by the 2 operators were less than 3 μm in all parameters evaluated.
|Measurement 1||Measurement 2||Measurement 3||P Value a||Measurement 4||P Value b|
|Average RNFL, μm||99.53 ± 10.10||98.68 ± 10.80||99.13 ± 9.98||.34||98.59 ± 10.04||.03|
|Superior RNFL, μm||125.57 ± 18.05||123.66 ± 18.10||125.11 ± 17.20||.64||124.36 ± 16.98||.30|
|Temporal RNFL, μm||68.78 ± 10.68||68.94 ± 11.10||68.61 ± 9.21||.34||68.41 ± 10.10||.56|
|Inferior RNFL, μm||133.14 ± 17.34||131.40 ± 20.14||130.82 ± 21.12||.39||131.56 ± 16.47||.08|
|Nasal RNFL, μm||70.99 ± 12.02||70.74 ± 13.01||72.19 ± 12.69||.68||70.28 ± 13.40||.39|
|Rim area, mm 2||1.59 ± 0.33||1.59 ± 0.31||1.62 ± 0.32||.77||1.57 ± 0.32||.48|
|Disc area, mm 2||2.06 ± 0.38||2.05 ± 0.37||2.09 ± 0.36||.82||2.02 ± 0.38||.18|
|Cup-to-disc area ratio||0.43 ± 0.17||0.42 ± 0.17||0.43 ± 0.16||.94||0.42 ± 0.17||.21|
|Macular thickness, μm||282.91 ± 11.83||282.76 ± 12.04||282.56 ± 12.59||.67||282.29 ± 10.16||.14|
|Macular volume, mm 3||10.18 ± 0.43||10.18 ± 0.44||10.17 ± 0.46||.52||10.16 ± 0.44||.15|
One hundred children were included in the intraobserver study. All RNFL and macular measurements were highly reproducible. COVs and ICCs of the measurements are presented in Table 2 . Optic nerve parameters showed lower reproducibility. The parameter exhibiting the highest intraobserver reproducibility was the average RNFL with a COV of 2.24%. ICCs showed an almost perfect reliability in all the parameters evaluated (ICCs > 0.81), and the ICC was 0.957 for the average RNFL. Figure 1 shows Bland-Altman plots of the average RNFL thickness and macular thickness reproducibility between the intraobserver measurements.
|Cup-to-disc area ratio||11.25||0.972|