An axial corneal topographical mapof a right eye at birth (A) and at 3 months (B). The central corneal powerdecreased from 49.1 to 45.1 diopters (D), and the astigmatism diminished from5.7 D at 96° to 2.5 D at 111°.
An axial corneal topographical mapof one of the steeper left eye corneas at birth (A) and at 6 months (B). Thecentral corneal power decreased from 54.9 to 47.3 diopters (D), and the astigmatismdiminished from 8.1 D at 116° to 0.1 D at 5°.
Isenberg SJ, Del Signore M, Chen A, Wei J, Christenson PD. Corneal Topography of Neonates and Infants. Arch Ophthalmol. 2004;122(12):1767-1771. doi:10.1001/archopht.122.12.1767
To evaluate corneal curvature by direct topographic analysis duringthe first 6 months of life.
We evaluated corneal topography in 200 infants using a specialized handheldtopographic instrument at a mean of 1.6 days after birth, and in some againat 3 and 6 months in the newborn nursery and ophthalmology clinic of a publichospital.
At birth, the mean central corneal power measured 48.5 diopters (D)(95% confidence interval [CI], 48.2-48.8 D; range, 41.4-56.0 D) and astigmatismmeasured 6.0 D (95% CI, 5.6-6.3 D), usually “with the rule” (80%)with a mean axis of 95°. The mean astigmatism on the semimeridian mapat 3 mm was 6.4 D (95% CI, 6.0-6.8 D); and at 5 mm, 5.9 D (95% CI, 5.4-6.3D). At birth, neonates delivered vaginally had a greater frequency of with-the-ruleastigmatism than those delivered by cesarean section (P = .02). By 6 months, the mean central corneal power andastigmatism decreased to 43.0 (95% CI, 41.3-43.1) D and 2.3 (95% CI, 1.4-3.2)D, respectively (P<.005 for each).
Newborns have steep, high, astigmatic (generally with-the-rule) corneasat birth that flatten significantly by the age of 6 months. The method ofdelivery can affect the astigmatic axis at birth.
The shape of the corneal surface in infants has been quite steep inseveral studies. Clinicians have appreciated this steepness when donor corneasfrom infants were transplanted onto adult recipients, resulting in keratometrymeasurements averaging 57.7 diopters (D).1
Few studies have attempted to directly measure corneal curvature inneonates and infants. Most had a small sample size and used techniques todetermine the refractive error and then secondarily implied the contributionof the cornea. Others have used keratometers. Keratometers, however, assumea spherocylindrical corneal shape and are limited to monitoring the cornealcurvature based on 4 points. Placido disc–based videokeratography, asused in this study, has provided reliable data about the anterior cornealsurface, approaching the spherocylindrical ideal.2,3
To our knowledge, this study is the first to evaluate corneal curvatureby direct topographic analysis in neonates and infants, as determined by asearch on the PubMed Web site of the National Library of Medicine and in ourliterature review.
The Human Subjects Committee of the Harbor-UCLA Medical Center approvedthe protocol. Written informed consent was obtained from each mother beforeher child was studied. The infants were securely wrapped in a swaddling cloth.The eyelids were separated with 2 cotton-tipped applicators and were alwayselevated off the ocular surface. Great care was taken to apply the applicatorsonly to the upper and lower rims of the orbit. No pressure was applied directlyon the eye.
The topography of each cornea was then studied with a specialized topographicinstrument (Vista Handheld Corneal Topographer; EyeSys Premier, Irvine, Calif),a Placido disc–based system. Before each use, the instrument was calibratedaccording to the manufacturer’s instructions. A study was accepted onlyif at least 3 complete Placido rings were visible, centered, and properlyfocused on the recorded image. If these criteria were not met, the study wasrepeated until the criteria were satisfied. The data were directly downloadedto a portable computer and analyzed with software for the topographic instrument.
Data for each eye of each infant, where available, were used in allstatistical analyses, summarized with means and frequencies over all eyes.Confidence intervals (CIs) and statistical comparisons that incorporate thecorrelation between the 2 eyes of an individual were used to include all availabledata without inflating the precision of estimates. Linear mixed models withcompound symmetric correlation were used for corneal power and astigmatism.4,5 Generalized estimating equations witha binomial distribution and a logit-link function were used to compare “with-the-rule”astigmatism between vaginal and cesarean section deliveries, incorporatingwithin-subject correlation between eyes.6 Analyseswere performed with modules (Mixed and Genmod) in SAS statistical software,version 8.2.7 To verify the validity of thetopographic system, we compared the keratometric data of 22 adult eyes, asmeasured by the same system, with measurements obtained from 2 other instruments(IOLMaster [Carl Zeiss, Jena, Germany] and ARK-760A [Marco, Jacksonville,Fla]).
Of the 200 neonates studied, 95 were female. All newborns who met thecriteria and whose parent consented were consecutively recruited for the studyfrom March 7, 1999, through June 6, 2000. The racial proportion was as follows:74% Latin American, 12% African American, 8% white, and 5% other (percentagesdo not total 100 because of rounding). The mean ± SD birthweight was 3318 ± 543 g (range, 1890-5030 g). At examination,the mean ± SD postnatal age was 1.6 ± 1.2days (range, 0-8 days), and the mean ± SD postconceptionalage was 39.3 ± 1.7 weeks (range, 29.9-42.7 weeks). Repeatexaminations were conducted at a mean ± SD of 3.0 ± 0.5and 6.0 ± 0.5 postnatal months.
For the neonates, the mean central corneal power was 48.5 (range, 41.4-56.0)D. The mean keratometric astigmatism was 6.0 (range, 0.2-16.4) D. The meanastigmatism on the semimeridian map at 3 mm was 6.4 (range, 0.01-22.5) D;and at 5 mm, 5.9 (range, 1.1-18.3) D.
The mean axis of the astigmatism in the neonates was 95.0°. Mostof the astigmatic axes (80%) were with the rule (within 20° of 90°),while 19% were at an oblique axis and less than 1% were “against therule” (within 20° of 180°).
Neonates delivered vaginally had the same (P = .49)mean central corneal power (48.5 D; 95% CI, 48.2-48.9 D) as those deliveredby cesarean section (48.5 D; 95% CI, 47.7-48.9 D). The astigmatism was alsothe same (P = .47): 6.0 (95% CI, 5.6-6.4)D and 6.0 (95% CI, 5.0-6.4) D, respectively. By 3 months, the difference forcentral corneal power was 44.5 (95% CI, 43.4-45.6) D and 44.1 (95% CI, 42.7-45.3)D, respectively, and the astigmatism was 3.6 (95% CI, 2.6-4.5) D and 2.9 (95%CI, 1.6-3.9) D, respectively; these values were also not statistically different(P>.30).
However, the type of delivery was related to the astigmatic axis. Atbirth, neonates delivered vaginally had a greater frequency of with-the-ruleastigmatism (83%) than those delivered by cesarean section (72%) (P = .02). By 3 months, the frequencies were similar (86%and 73%, respectively) to those at birth, but nonsignificant (P = .37) because of the reduced number of infants observedat this point. There are insufficient data for comparison at 6 months.
The follow-up data for the 3- and 6-month examinations are given inthe Table. Figure 1 and Figure 2 showexamples of differential maps from birth to the ages of 3 and 6 months, respectively.The decreases in central power and astigmatism from birth to the 3- and 6-monthmeasurements were statistically significant (P<.005).From 3 to 6 months, the reductions in central power and astigmatism were notsignificant (P = .93 and P = .14, respectively). Frequencies of type of astigmatismare almost identical at birth and at 3 months. An observed reduction in frequencyof with-the-rule astigmatism at 6 months is nonsignificant (P>.10).
For all comparisons at all ages studied, there were no statisticallysignificant differences whether the infants were stratified by sex or race(P>.05 for all).
The validation studies on the adult eyes found no significant differencesamong the instruments (P>.70). For steepest and flattestmeridians compared with a different instrument (IOLMaster), the topographicsystem used (Vista) demonstrated a mean difference of 0.13 D flatter and 0.03D steeper. The numbers for the comparison with the second instrument (ARK-760A)were 0.1 D steeper and 0.2 D steeper.
There have been few studies that directly measured corneal curvaturein newborns. In 1976, Ehlers and colleagues8 reportedthe mean ± SD corneal curvature in 19 mature newborns tobe 7.1 ± 0.07 mm using a keratometer. By using an ophthalmometeron neonates, Gordon and Donzis9 found the mean ± SDapical keratometry to decrease from 53.6 ± 2.5 D in 11 eyesat 30 to 35 postconceptional weeks to 51.2 ± 1.1 D in 10eyes at 39 to 41 postconceptional weeks. In 11 infants younger than 1 year,they found the mean ± SD keratometric value to further decreaseto 45.2 ± 1.3 D. There was no measurement of astigmatism.While these numbers are higher than our corresponding numbers of 48.5 D atbirth and 43.0 D at 6 months, their measurements are within 1 SD of our findings.Interestingly, the nearly 6-D reduction they found from birth to 1 year isquite similar to the reduction from birth to 6 months reported in this study.In Japan, Inagaki10 performed automated keratometryon 11 mature neonates and found a mean ± SD keratometry readingof 47.0 ± 1.2 D. The curvature decreased to a mean ± SDof 44.1 ± 1.7 D by the age of 3 months in the 8 infants hefollowed up. These numbers are close to the results of this investigation.A 2004 study, which used a keratometer, reported higher values than thosefound in this study—58.6 D horizontally and 54.0 D vertically when averagingboth eyes.11
The use of infantile corneas for keratoplasty to adult recipients hasbeen reported. Koenig and associates1 believedthat the large value of 57.7 D following such a keratoplasty might reflecta steepening effect induced by suturing the relatively elastic infantile corneato an adult recipient. Wood and Nissenkorn12 attributedthe myopia found in their adult recipients of infant corneas to corneal steepness,with an average keratometry reading in the adults of 46.5/50.9 D.
In a 2004 study, Friling and associates,11 usingkeratometry, reported 6 D of cylinder in their 32 infants whose postconceptualage was younger than 32 weeks and 3.3 D for the 30 infants whose postconceptualage was older than 36 weeks.
The nature and amount of astigmatism reported in neonates and infantsare controversial. By using near retinoscopy without cycloplegia, Mohindraand colleagues13 found that, of 17 newbornsyounger than 1 week, about 20% had astigmatism of 2 D or more. The same 20%proportion was still present at the age of 6 months. In a later report, astigmatismof at least 1 D in 55% of infants younger than 5 months was found, with 10%displaying a cylinder exceeding 3 D.14
By using photorefractive techniques, Howland and associates15 studied 15 neonates younger than 10 days, and reportedthat 50% had astigmatism, with 25% exceeding 1 D. At the age of 6 months,the proportion was also parallel. With similar methods, Atkinson and coworkers16 found that almost all of their 20 infants at theage of 3 months had at least 1 D of astigmatism and that the quantity wasreduced to adult levels by the age of 18 months. Weale17 calculatedthat the with-the-rule astigmatism should measure 3.5 D in white and 5.8 Din Japanese neonates.
We measured a greater mean astigmatism at birth (5.6 D) and 3 months(3.3 D) than these previous studies. This could be attributed to the strongpossibility that the direct topographical analysis used in this study is moreaccurate than the other methods attempted. To ensure that the handheld systemwe used was as accurate as other instruments, we compared it with 2 otherdevices and found the measurements nearly identical among the 3. It is alsopossible that we measured the neonates closer to birth (mean, 1.6 days, withsome as early as a few hours after birth) than the other studies. Thus, wewould more likely appreciate the direct effects of the birth process, whichmay have compressed the eye and produced a larger astigmatism than reportedin previous studies. By the age of 6 months, our findings (2.3 D) are moresimilar to previously reported values.
Most cesarean section deliveries were begun after the neonate’shead was already engaged in the birth canal. This would explain our findingthat a vaginal delivery caused a similar amount of astigmatism as a cesareandelivery.
Previous reports of the axis of astigmatism in newborns and infantsare even more contradictory. By using near retinoscopy without cycloplegia,Gwiazda and coworkers14 examined 440 infantsyounger than 5 months. They found that the astigmatism of at least 1 D thatwas present in 235 infants was with the rule in 45% of the cases and againstthe rule in 45%. This study was somewhat contradicted by a subsequent investigationby Thorn and colleagues.18 Also using nearretinoscopy without cycloplegia, Thorn et al reported that 51% of 45 whiteinfants (aged 3½-11.8 months) had with-the-rule astigmatism and only13% had against-the-rule astigmatism. They also reported that 59% of 22 Chineseinfants (aged 3½-12 months) had against-the-rule astigmatism.
There are potential problems with studies using near retinoscopy. Wessonand colleagues19 found significant differencesbetween near retinoscopy and cycloplegic refraction for sphere and cylinderpower, especially in infants. Maino and coworkers20 foundnear retinoscopy and cycloplegic retinoscopy to agree (within 0.5 D) in only27% of 311 children between the ages of 18 and 48 months. Saunders21 indicated several potential sources for error usingthis technique, including not neutralizing the meridians simultaneously whencycloplegia is not used and performing retinoscopy off axis, especially inthe highly curved infant cornea. She also was concerned about the large disparityin the angle between the visual and optical axes (angle α) in newborns(8°-10°) compared with adults (4°-5°). This angle αeffect might influence not only retinoscopy but also techniques using camerasfor photorefraction. Indeed, using isotropic and orthogonal photorefractivetechniques, Howland and Sayles22 reported anagainst-the-rule astigmatism rate of about 50% in 76 astigmatic infants youngerthan 1 year (mean age, 6 months).
In a recent study, Mayer and colleagues23 reportedthat all their 1-month-old infants with greater than 1 D of cylinder demonstrateda with-the-rule axis. By the age of 6 months, the proportion had decreasedto about a third. Weale17 postulated that thewith-the-rule astigmatism and corneal ellipticity found in newborns are attributableto the oblate form of the fetal eye, supported by the observation that theyoung cornea yields less to stress than does the sclera.
Our findings of with-the-rule astigmatism in about 80% of infants atbirth and at the age of 3 months show a greater preponderance of with-the-ruleastigmatism than previous studies. Possible explanations for this discrepancymight include the increased accuracy implicit in this study by using directcorneal topography instead of somewhat subjective retinoscopy or photographictechniques. Our observation of the with-the-rule frequency decreasing to 50%by the age of 6 months, while the oblique and against-the-rule groups wereincreasing (which is, however, based on only 8 infants) is consistent withthe general trend over life to shift from with-the-rule to against-the-ruleastigmatism.24
The type of delivery was related to the astigmatic axis. At birth, with-the-ruleastigmatism was more frequent in neonates delivered vaginally (83%) than inthose delivered by cesarean section (72%) (P = .02).Because cesarean sections were generally begun shortly after the head wasengaged in the birth canal, the angle of compression on the eye may have beendifferent than following a complete vaginal delivery. A similar differenceaccording to type of delivery was observed at 3 months, but is not statisticallysignificant because fewer infants were observed at this point.
This study does have limitations. To open the eyelids, we used cotton-tippedapplicators. To minimize the possibility that pressure was exerted on theglobe, we applied direct pressure only to the orbital rims and actually generallyelevated the eyelids off of the ocular surface. We chose this method becauseprevious studies found problems using an eyelid speculum. Masci25 demonstratedthat use of a speculum reduced the cylinder in with-the-rule subjects by amean ± SD of 0.7 ± 0.2 D, and increasedthe cylinder in against-the-rule subjects by a mean ± SDof 0.2 ± 0.2 D. Wilson and colleagues26 foundthat, for eyes with toricity greater than 0.6 D with the rule, the mean shiftinduced by an eyelid speculum was −0.1 D vertically and −0.4 Dhorizontally. Shea and associates27 recentlyreported that the use of eyelid specula affects the measured cycloplegic refractionwhen children are under anesthesia. For children younger than 4 years, themeasured cylinder had an average difference of 1.7 D when comparing 2 differenttypes of eyelid specula (P<.001). Recently, theuse of an eyelid speculum in children was shown to increase the intraocularpressure an average of 4 mm Hg.28 To avoidthe artifact induced by using a speculum, we decided to use a manual technique.It is possible that despite our efforts, some pressure may have contactedthe globe in some infants. We also lost several infants for the 3- and 6-monthfollow-up examinations. These were healthy infants with little reason to returnfor the examination, despite our efforts to contact all of them.
In conclusion, corneal topography determinations in neonates revealsteep, high, astigmatic (generally, with-the-rule) measurements at birth thatflatten significantly by the age of 6 months. The method of delivery affectsthe astigmatic axis at birth, but not the amount of astigmatic cylinder.
Correspondence: Sherwin J. Isenberg, MD,Department of Ophthalmology, Harbor-UCLA Medical Center, 1000 W Carson St,Torrance, CA 90509 (email@example.com).
Submitted for Publication: November 14, 2003;final revision received April 12, 2004; accepted June 1, 2004.
Funding/Support: This study was supported inpart by General Clinical Research Center grant MO01 RR00425 from the NationalCenter for Research Resources, Bethesda, Md; the Sara Kolb Fund, Los Angeles,Calif; the Kirchgessner Foundation, Los Angeles; and a Senior Scientific InvestigatorAward from Research to Prevent Blindness, New York, NY (Dr Isenberg).
Financial Disclosure: None.
Acknowledgment: We thank Robert Maloney, MD,for his advice and review of the manuscript; and Nancy Berman, PhD, for herstatistical services.