Fast macular thickness map. A, Copy of printout as generated by the optical coherence tomography (OCT)–3 machine (Carl Zeiss Meditec, Dublin, California). B, Linear regression of average outer macular thickness (OMT) as measured by OCT-3 vs axial length (AL) in healthy black and white children, separately. For black children, R2 = 0.0507, P = .24. For white children, R2 = 0.2014, P = .003. C, Linear regression of foveal thickness (FT) as measured by OCT-3 vs age in healthy black and white children, separately. For black children, R2 = 0.068, P = .001. For white children, R2 = 0.0034, P = .89.
Fast retinal nerve fiber layer (RNFL) thickness scan (3.4-mm circle protocol). A, Copy of printout generated by the optical coherence tomography (OCT)–3 machine. B, Linear regression of average RNFL thickness as measured by the OCT-3 vs axial length (AL) in healthy black and white children, separately. For black children, R2 = 0.0006; P = .14. For white children, R2 = 0.09; P < .001. S indicates superior; T, temporal; N, nasal; and I, inferior.
Fast optic disc protocol. Copy of the figure part of the printout generated by the optical coherence tomography (OCT)–3 machine. Both children had healthy eyes. Notice the deeper cup of the black child. S indicates superior; T, temporal; N, nasal; and I, inferior.
El-Dairi MA, Asrani SG, Enyedi LB, Freedman SF. Optical Coherence Tomography in the Eyes of Normal Children. Arch Ophthalmol. 2009;127(1):50-58. doi:10.1001/archophthalmol.2008.553
To collect a normative database of macular thickness, retinal nerve fiber layer (RNFL) thicknesses, and optic nerve topography in the healthy eyes of children aged 3 to 17 years using optical coherence tomography (OCT) measurements.
Scans were obtained for 286 healthy children (black, 114; white, 154; other, 18). Each child had a dilated eye examination, an axial length measurement using the IOL Master (Carl Zeiss Meditec, Dublin, California), and OCT measurements using the fast macular map, fast RNFL thickness, and fast optic disc protocols of the Stratus OCT (OCT-3; Carl Zeiss Meditec).
Black children had smaller macular volume and foveal thickness, larger RNFL thickness, and larger cup-disc area ratios compared with white children. Macular volume and average outer macular thickness correlated negatively with axial length in white children. Foveal thickness correlated positively with age in black children only. Average RNFL correlated negatively with axial length in white children only (P < .05 for all). Normative data for all variables were recorded and compared with reported adult values.
Stratus OCT–3 measurements of macular and RNFL thickness and optic nerve topography vary with race, axial length, and age in healthy children. Normative pediatric OCT data should facilitate the use of OCT in assessing childhood glaucoma and other diseases.
Children with glaucoma or any other type of progressive optic neuropathy are more difficult to diagnose and to monitor for disease progression than adults because of the challenge in obtaining reliable and reproducible visual fields and intraocular pressures. Optical coherence tomography (OCT) is used in patients with glaucoma to measure the retinal nerve fiber layer (RNFL) and macular thickness. It has been shown to have good sensitivity to detect glaucoma in adults1,2 and is fairly easy to use in children.3- 10 Optical coherence tomography is based on the principle of Michelson interferometry and uses low coherence light, which is reflected by retinal tissues. The technique is noninvasive and well tolerated.11,12 The Stratus OCT (OCT-3; Carl Zeiss Meditec, Dublin, California) machine has an integrated normative database; however, it only included individuals aged 18 years or older. The few published reports of OCT values in the healthy eyes of children are limited by consideration of either a single age group3,13- 16 or a single race.3,10 Previous studies have shown that OCT values are affected by age,2,17,18 axial length,19- 21 and race.22,23
The purpose of this study is to collect normative pediatric values for OCT measurement of the macular thickness and volume, peripapillary RNFL thickness, and optic nerve head morphology, and to study the effects of age, race, axial length, and spherical equivalent on these values.
From November 2006 until November 2007, 286 children aged 3 to 17 years were enrolled from the Duke Eye Center pediatric clinic. Included were healthy full-term children with no known ocular pathology other than childhood strabismus or treated uniocular amblyopia with a final visual acuity of better than 20/30. The study was approved by the Duke University institutional review board and was in accordance with Health Insurance Portability and Accountability Act. All subjects underwent a complete eye examination by a fellowship trained pediatric ophthalmologist (M.A.E., L.B.E, and S.F.F.). Informed consent was obtained from the child's legal guardian. Data gathered included age, sex, parent-reported race, ocular and medical history, cup-disc ratio, cycloplegic refraction, OCT measurements, and axial length measurement.
Eyes of subjects with high refractive errors defined as a spherical equivalent of more than ±6.0 diopters (D) or astigmatism more than 3.5 D as well as eyes with a clinically determined cup-to-disc ratio of more than 0.5 were excluded.
Axial length (AL) measurements were obtained using the IOL Master (Carl Zeiss Meditec). Noncontact axial length measurements were performed 3 times, and the average of the 3 values was recorded. Poor signal values as well as values that differed by more than 0.1 mm were rejected and the measurement was repeated.
Optical coherence tomography scans were performed using the Stratus OCT. All scans used an internal fixation target and a default eye length of 24.46 mm. Scans with signal strength of less than 6 or showing any artifacts on the individual scans were rejected and then repeated. All OCT scans were performed by the same operator (M.A.E.). The right eye in subjects with equal vision or the better-seeing eye in subjects with unilateral amblyopia was tested first. Three protocols were performed for each eye of every subject in the following order: (1) The fast macular thickness map (Figure 1A) consists of 6 radial scans centered on the fovea, with each scan formed by 128 single A scans. The machine displays the thickness values in 3 rings. The foveal area corresponds to the innermost 1-mm diameter, the inner ring to 3-mm diameter, and the outer ring to 6-mm diameter. The macular volume for each eye is also reported. (2) The fast RNFL thickness scan (3.4-mm circle protocol) (Figure 2A) consists of 256 individual A scans in a path around the circumference of a circle 3.46 mm in diameter and centered around the optic nerve. The machine scans the area 3 times consecutively and gives the average thickness value of the 3 scans for each clock hour and 4 quadrants as well as an overall average thickness value for the entire circumference. (3) The fast optic disc protocol (Figure 3) gives a topographic map of the optic nerve and cup, giving an optic disc area, cup area, and rim area value. In this protocol, when a clinical cup-disc ratio is 0 by the examining physician, the OCT machine interprets it as 1. This error was corrected manually by the operator before logging the data.
Analysis included the right eyes of binocular children and the nonamblyopic eyes of children with uniocular amblyopia or a history of treated amblyopia. Data were analyzed using Microsoft Excel (Redmond, Washington) spreadsheets as well as JMP 6.0 (SAS Institute, Cary, North Carolina). A Shapiro-Wilk W test was performed to test for normality of the data. For multivariate analysis, a generalized linear model was made for each of the OCT variables, and the effects of race, sex, age, spherical equivalent, and axial length were tested. Data were analyzed for the entire population initially and then for the black and white children separately because those 2 racial groups had the largest representation in our sample. Correlation coefficients for each factor were calculated, as well as analyses of variance and P values. Unpaired t tests were used to compare variations between the white and black children in the whole sample as well as in 3 age groups (group 1, 3-6 years; group 2, 7-10 years; group 3, 11-17 years).
All P values were Bonferroni adjusted, with P values less than .05 considered statistically significant.
To adjust the RNFL thickness for axial length variation in a study of Turkish adults, Bayraktar et al19 developed a statistically derived formula, RNFLestimated integral (RNFLEI), that theoretically corrects OCT measurements of RNFL for the variation in axial length. The RNFLEI value is calculated as RNFLEI = C × RNFL thickness, where C indicates magnification correction and is calculated as 10.87/C = [1 + (0.018 × Daxial) + (0.002 × Drefraction)], where D indicates refractive power.
In the present study, RNFLEI was calculated and the generalized linear model was run initially for the whole sample, then for black and white children independently.15,19
Of 350 eligible children, 296 consented to the study (84.5%). Ten were excluded because of inability to obtain a good quality macular scan. The values for each OCT measurement were found to be normally distributed (data not shown). As reported by the children's parents, there were 114 black, 154 white, and 18 children of mixed or other races. The mean (SD) age was 8.59 (3.11) years; there were 85 children in age group 1, 120 in group 2, and 79 in group 3.
Black children were slightly less hyperopic than white children (spherical equivalent for black children, 0.27 D; white, 0.77 D; P = .03), although the axial lengths were equivalent.
Results of this scan are summarized in eTable 1, eTable 2, (www.archophthalmol.com) and Table 1. Values of macular thickness and volume as measured by OCT are reported for all children and then for black and white children separately (eTable 1). The inner macular thickness and foveal thickness were both found to be greater in white than in black children (P < .001 for both). The total macular volume was also greater in white vs black children (P = .03) (eTable 1). These racial differences were more significant in the 2 younger age groups (eTable 2). In the oldest age group, only the fovea was significantly thicker in white children than in black (P = .02) (eTable 2).
Multivariate analysis (eTable 3) showed significant negative correlation of the outer macular thickness values with axial length in white children (Figure 1B) and a positive correlation of both the inner macular thickness values and the foveal thickness with age in black children (Figure 1C). The average outer macular thickness decreased by about 5 μm with each 1-mm increase in axial length in white children (P = .003), and foveal thickness increased by about 1.7 μm for every 1-year increase in age in black children (P = .02).
The distribution of macular thickness values in micrometers and macular volume values in cubic millimeters by age group and race is shown in Table 1.
These are illustrated in eTable 4, eTable 4, and Table 2. Values of RNFL thickness as measured by OCT were reported for all children, then for the black and white children separately (eTable 4). The superior quadrant and average RNFL thickness were significantly greater in black vs white children (P < .001 for both). This difference in superior RNFL thickness between black and white children was present across all 3 age groups (eTable 5).
Multivariable analysis () showed a negative correlation of the average RNFL thickness with both age and axial length in white children only. The average RNFL thickness decreased by about 2.6 μm with every 1-mm increase in axial length in white children (P < .001). Values of RNFL thickness were independent of all variables tested in black children (Figure 2B).
The distribution of RNFL thickness values by age group and race is shown in Table 2.
The calculated RNFLEI values were independent of axial length and age in white children. In contrast, RNFLEI showed a positive correlation with axial length in black children.
These scans are illustrated in eTable 6, eTable 7, and Table 3. Values of the optic disc, cup, and cup-disc areas as well as horizontal and vertical cup ratios as measured by OCT were reported for all children, then for the black and the white children separately (eTable 6). Black children had a larger average cup area and cup-disc ratio than white children (P = .01 and .002, respectively). On subgroup analysis by age, this difference was statistically significant only in the oldest age group (eTable 7).
Multivariate analysis (eTable 3) (P < .001) showed a positive correlation of the cup area with axial length in the whole population. Using subgroup analysis, this relationship was statistically significant among black but not among white children.
The distribution of nerve morphology values by age group and race is shown in Table 3.
This study reports normative values for OCT measurements of macular thickness and volume, peripapillary retinal nerve fiber layer thickness, and optic nerve morphology in children. Differences in OCT measurements between the healthy eyes of black and white children were examined for effects of age, axial length, and refractive error.
Several articles have reported OCT measurements in normal children. The Sydney Childhood Eye Study13- 16,24 reported OCT data from 1765 six-year-old children selected from schools. Most were white (909 children) and East Asian (213 children); the study included eyes with large cup-disc ratios (up to 0.9), and was limited to a single age group. Salchow et al10 studied the RNFL thickness in 92 eyes of children aged 4 to 17 years; 91% of the children were Hispanic. Ahn et al3 reported RNFL thickness values in 72 Korean adolescents. By contrast, pediatric OCT values reported by Gupta et al24 differed from all other reports, possibly owing to technical differences in the protocols used. The authors in this small study did not specify which generation of OCT machine they used.
Table 4 compares our OCT values with those of the above articles as well as relevant adult studies. We are unaware of any previous reports of normal OCT values for black children. Average RNFL thickness values for our total sample were comparable with those of Salchow et al10 and Ahn et al3 (population largely Hispanic and all Korean, respectively) but higher than those of the younger subjects reported by Parikh et al18 (population mostly Asian Indian). Compared with the large study of Australian 6-year-olds by Hyunh et al,15,16 both macular volume and average RNFL thickness were higher in our study population. Possible explanations for this difference include our stricter optic nerve and refractive criteria, and the existence of variation between Caucasian subjects in distinct geographic areas.
Compared with adult studies, average RNFL thickness in our population was comparable with but slightly higher than that reported in the younger adults (age, <40 years) by Budenz et al25 and Leung et al.26 By contrast, the average RNFL thickness in our study was slightly lower than that reported by Racette et al23; this author found higher RNFL thickness values in both black and white adult subpopulations compared with most other published adult studies, probably owing to the study's use of a different version of the OCT software.1,18,26
We found differences in OCT measurements between black and white children at the levels of the macula, RNFL thickness, and optic disc topographic parameters. These differences were also found to vary among age groups. Black children in the present study had thinner foveal and inner macular thicknesses and higher RNFL thickness values than their white counterparts. Older black children also had a larger cup areas and cup-disc ratios than white children of similar age, although the neuroretinal rim area was similar between the 2 racial groups.
Variations in the optic nerve morphology among different races have been shown previously in adults using optic nerve stereophotographs and special imaging.23,27 Differences were also shown on stereophotographs between black and white children28 and on OCT between white and East Asian children,15 both at the level of the optic nerve and the RNFL, but not the macula.16
Racette et al23 have shown, using confocal scanning laser ophthalmoscopy, that normal black adults have larger optic nerves, larger cup-disc ratios, larger cup volumes, and similar neuroretinal rim area than their white counterparts. Poinoosawmy et al29 and Tjon-Fo-Sang et al30 found larger mean RNFL thickness in black subjects compared with white ones using scanning laser polarimetry. Racette et al23 also found, using OCT, that normal black adults have larger mean RNFL thickness compared with white adults.23
We found 2 articles that compared macular values by OCT in white and black adults. Hence, Asefzadeh et al22 (14 subjects) reported lower foveal and total macular thickness values in black adults than in white adults. Kelty et al31 (83 subjects) reported lower foveal thickness as well as lower values in 3 of 4 areas of the inner macular thickness in black adults compared with white ones. Our findings in children are consistent with data reported in these adult studies.
The OCT measurements of RNFL thickness have been previously described as dependent on race, age, axial length, and disc area in adult eyes,25,32 with a reported average RNFL thickness decrease of 2.2 μm for every 1-mm increase in axial length, and an average RNFL thickness decrease of 2 μm for every decade of aging.15
While we found a similar relationship between RNFL thickness and axial length measurements in white children, RNFL thickness was independent of axial length in black children. Bayraktar et al19 derived a statistical formula designed to correct measured RNFL thickness for axial length in Turkish adults (see Methods). When we applied the formula to our population, the calculated RNFLEI was found to be independent of axial length in white children but showed a new positive correlation with axial length in black children, suggesting that this formula may not yet be widely applicable.
Measurements of RNFL thickness were not dependent on age in our population of children younger than 18 years. This finding is not surprising, given that published studies of RNFL thickness in adults of different age groups do not demonstrate much reduction of the average RNFL thickness until 40 to 60 years of age (Table 4).18,25 These findings are consistent with pathology studies by Dolman et al33 suggesting that change due to age-related decay of the optic nerves is not significant until after age 60 years.
Measurements with OCT of the inner and outer macular thickness in children have been reported to decrease with both increasing axial length and myopia.16 We found a similar relationship between outer macular thickness and axial length in our study population, which was also statistically significant within the white but not the black subgroups. Interestingly, we also noted an increase in foveal and inner macular thickness with age in black children, a finding previously unreported to the best of our knowledge (PubMed search parameters were “normal foveal thickness,” “optical coherence tomography,” and “from 1950 until the present”).
Samarawickrama et al20 have shown that OCT measurements of the cup area positively correlate with axial length in children. We found a similar relationship in our population; however, on subgroup analysis by race, this relationship was present only in the black subpopulation. This result may not have shown for our white subjects owing to the exclusion of large refractive errors, and therefore extremes of axial length, in our study.
Limitations of the present study are several. Limited numbers and the exclusion of eyes with high refractive errors may have limited our ability to identify additional relationships between macular thickness and RNFL thickness and axial length. Owing to the cross-sectional (rather than the longitudinal) nature of the study, we are unable to determine changes in OCT measurements over time in a given child's eye. In addition, though we were able to compare findings among white and black children, we did not have sufficient numbers of children belonging to other ethnic groups to comment on other ethnicity-related differences.
To reduce the chances of including children with glaucoma or other optic nerve abnormalities in this normative database, we deliberately set strict inclusion criteria regarding an acceptable cup-disc ratio. Hence we may have excluded children with physiologic optic nerve cupping. Because black children may physiologically have a larger cup-disc ratio than their white counterparts,27,28 we may have minimized the racial differences in OCT measurements within this study, specifically those related to OCT-measured optic disc morphology. In addition, by excluding children with high refractive errors from the present study, we have necessarily limited our normative data set.
This study has generated normative data for OCT measurement of macular thickness and volume, RNFL thickness, and optic nerve morphology in the healthy eyes of children (Tables 1, 2, and 3). This information should facilitate evaluation of OCT assessments performed in children with diagnosed or suspected glaucoma as well as those with other optic neuropathies.
Correspondence: Sharon F. Freedman, MD, Department of Ophthalmology, Duke University Eye Center, Durham, NC 27710 (firstname.lastname@example.org).
Submitted for Publication: July 7, 2008; final revision received August 26, 2008; accepted September 3, 2008.
Funding/Support: An OCT-3 machine was loaned for data acquisition by Carl-Zeiss Meditech, Dublin, California. Dr Asrani is a recipient of a Career Development Award from Research to Prevent Blindness.
Previous Presentations: This work was presented in part at the Annual Meeting of the American Association of Pediatric Ophthalmology and Strabismus; April 5, 2008; Washington, DC.
Financial Disclosure: None reported.