A

Ocular Growth and Refractive Error Development in Premature Infants with or without Retinopathy of Prematurity
Anne Cook,1 Sarah White,2 Mark Batterbury,3 and David Clark4
PURPOSE. To study factors involved in the development of refractive error in premature infants with or without retinopathy of prematurity (ROP). METHODS. Premature infants in the national ROP screening program were recruited and examined longitudinally between 32 and 52 weeks postmenstrual age. Axial length (AL), anterior chamber depth (ACD), and lens thickness (LT) were measured on the A-scan biometer. Corneal curvature was recorded with video-ophthalmophakometry and refractive state was determined with routine cycloplegic retinoscopy. Multilevel modeling techniques were used to study relationships between all the variables and stage of ROP throughout the study period, as well as individual growth rates. RESULTS. One hundred thirty-six infants were included. AL and ACD showed linear patterns of growth, whereas LT changed little over the study. Corneal curvature showed quadratic growth patterns in infants unaffected by ROP, but showed linear growth if ROP developed. Corneal curvature correlated well with refractive state. Most infants were myopic at the start of the study, became emmetropic around term, and were hypermetropic toward the end of the study. However, the eyes that were treated for ROP showed little change in refractive error; with signicantly less hypermetropia by the end of the study. CONCLUSIONS. Eyes of premature infants have shorter axial lengths, shallower anterior chambers, and more highly curved corneas than eyes of full-term infants. These differences become more signicant as the severity of ROP increases. Premature eyes develop less of the expected hypermetropia in fullterm eyes, mainly due to differences in ACD and corneal curvature. These differences are most signicant in eyes that receive laser treatment for ROP. (Invest Ophthalmol Vis Sci. 2008;49:5199 5207) DOI:10.1167/iovs.06-0114 yopia has been classied by the World Health Organization (WHO) as one of the leading causes of blindness and visual impairment in the world today.1 The prevalence of myopia varies, depending on age at examination, family history,2,3 ethnicity,2 and occupation.4 Environmental factors, genetic factors, premature birth, and the development of retinopathy of prematurity (ROP) are all known to be associated with the development of a specic form of myopia. Although there is a wealth of data on the development of myopia with ROP, few studies have prospectively documented changes in ocular growth along with alterations in refractive error during the earliest measurable weeks of premature life. Most studies commence after 3 months corrected age, and few prospectively record changes in all the biometric components and refractive error simultaneously. Our previous publication5 documented all such parameters in premature infants without ROP. The purpose of this study is to look prospectively at premature infants affected by ROP during the early phases of ocular growth and to identify factors contributing to refractive status at this time. It is hoped that comparing this group with a cohort without ROP who were studied at the same time may add to the existing knowledge of the factors that affect emmetropization after premature birth.

METHODS
This cohort of infants was recruited from the neonatal intensive care unit of Liverpool Womens Hospital. Regional ethics committee approval and parental consent were obtained. The infants examined included those falling within the screening criteria for the ROP screening program: infants born before 32 weeks gestational age and/or infants with birth weight below 1500 g. Any infant too unt for the examination necessary for the study was excluded. Infants were examined longitudinally at 32 (time point [T]1), 36 (T2), 40 (T3), 44 (T4), and 52 (T5) weeks postmenstrual age. These time points were deliberately chosen to be adequately spaced without interfering with or delaying the usual ROP examinations. Descriptions of the measurement methods follow. All methodology adhered to the tenets of the Declaration of Helsinki.

From the 1Manchester Royal Eye Hospital, Manchester, United Kingdom; the 2Division of Mental Health, University of London, London, United Kingdom; the 3St. Pauls Eye Unit, Royal Liverpool University Hospital, Liverpool, United Kingdom; and the 4University Hospital Aintree, Foundation Trust, Liverpool, United Kingdom. Supported by The Iris Fund for the Prevention of Blindness. Submitted for publication February 2, 2006; revised June 28 and November 30, 2006, April 18 and September 21, 2007, and January 14 and April 2, 2008; accepted October 21, 2008. Disclosure: A. Cook, None; S. White, None; M. Batterbury, None; D. Clark, None The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked advertisement in accordance with 18 U.S.C. 1734 solely to indicate this fact. Corresponding author: Anne Cook, Manchester Royal Eye Hospital, Oxford Road, Manchester, M13 9WH UK; cookydoc@hotmail.com
Investigative Ophthalmology & Visual Science, December 2008, Vol. 49, No. 12 Copyright Association for Research in Vision and Ophthalmology

Biometry
Axial length (AL), anterior chamber depth (ACD), and lens thickness (LT) were measured with an A-scan biometer (Carl Zeiss Meditec, Oberkochen, Germany), which was calibrated using the technique described by Butcher and OBrien.6 The method involves applanation of the cornea with the A-scan probe after the instillation of the topical anesthetic benoxinate hydrochloride 0.4%. The probe is placed lightly on the center of the cornea, perpendicular to its axis. It is maintained in this position until three clear traces are obtained on the screen. The average value from the three best images is recorded for all axial dimensions. (Posterior segment length [PSL] was calculated by sub-

5199

5200

Cook et al.

IOVS, December 2008, Vol. 49, No. 12

TABLE 2. Gestational Age and Range with Total Number of Infants Examined at Each Time Point Gestational Age (wk) Time Point T1 T2 T3 T4 T5 n 107 113 120 115 90 Mean 32.84 36.23 40.34 44.63 53.48 Range (wk) MinMax 30.7134.50 34.0038.86 36.0042.71 42.0055.00 48.0062.00

tracting the sum of ACD and LT from AL, and the result was checked with the printed scan.)

Corneal Curvature
The corneal curvature (CC) was measured with a video-based keratophakometer, previously described by Wood et al.7 The unit consists of a camera coupled to a video recorder. The camera has a Perspex faceplate that has illuminating infrared LEDs distributed around its perimeter. The mires are focused around the infants cornea and the image is captured on the video recorder for later analysis of CC. The camera was calibrated with a series of stainless-steel ball bearings of known radius of curvature at the start of each session.

Spherical Equivalent
Full cycloplegic retinoscopy was performed with a streak retinoscope, 30 minutes after the administration of 0.5% cyclopentolate and 2.5% phenylephrine. All refraction was performed by the same examiner (AC), with intermittent verication by the optometrist at University Hospital Aintree, to ensure accuracy. Handheld lenses were used to enable the examiner to ensure that the streak was kept on axis. The procedure proved to be relatively simple as, given the infants age, ocular movements were minimal during the refraction. An allowance of 1.5 D was made for a working distance of 23 m. Refractive error was recorded in the form of spherical equivalent (SE): SE Sphere Cyl/2.

Retinopathy of Prematurity
All infants were screened by the same examiner (DC) who has 16 years experience in screening premature infants. After instillation of an additional drop of benoxinate hydrochloride 0.4%, a lid speculum was placed gently between lids. Scleral indentation was then performed to allow examination of the far periphery. ROP grading was performed according to The International Classication of Retinopathy of Prematurity.8 All treatments were performed with a Diode laser.

correct for the lack of independence between observations. Many methods of analyzing longitudinal data require the same number of measurements to be collected from every subject (e.g., repeatedmeasures ANOVA), and for each subject to attend at every time period. In a clinical setting, this would have been unrealistic. Therefore this more appropriate analytical technique was used. Multilevel regression models (using PROC MIXED in SAS ver. 8.1; SAS, Cary, NC, for SunOS; Sun Microsystems, Santa Clara, CA) were used to look at how stage of ROP and postmenstrual age are related to AL, ACD, LT, CC, and refractive error (RE). The biometric parameters were regressed against postmenstrual age (weeks) to allow for the unequally spaced time intervals between examinations. Models were rst tted for infants at all stages of ROP. In this way, differences between stages of ROP could be examined, as well as a comparison of the growth rates between stages by including the interaction term of stage of ROP and postmenstrual age. The multilevel regression methodology allowed for the tting of a linear relationship or a quadratic relationship with postmenstrual age, depending on the growth curve exhibited. The method was decided on by testing the signicance of a quadratic term in each model to explore whether the parameter estimate for the quadratic term differed signicantly from 0 at the 5% signicance level, after the linear term was tted.

Statistical Analysis
Analysis of the refractive and biometric data from 44 pairs of eyes showed no signicant difference between the right and left eyes (t 0.722, P 0.474), with a mean difference 0.018 (95% CI, 0.033 0.069). On average, the difference in AL between pairs of eyes was only 0.018 mm. For this reason, the right eye of each infant was used for further data analysis. One-way analysis of variance was used to test whether stage of ROP was related to gestational age at birth and birth weight. Comparison of the stages of ROP at 3 months post term in term infants of the biometric parameters (AL, ACD, LT, and CC) was also made in this manner. Post hoc pair-wise comparisons were performed with Bonferroni correction. The relationship between stage of ROP, ocular growth, and refractive error was examined using multilevel modeling for repeated measures. This method was chosen as different observations in the same child are not independent, and simple regression analysis would not

TABLE 3. Ocular Parameters at Term (Intercept) According to Stage of ROP Parameter Axial length (mm) Stage 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 Intercept (SE) 16.66 (0.04) 16.51 (0.08) 16.45 (0.08) 16.63 (0.13) 16.37 (0.13) 2.26 (0.02) 2.24 (0.05) 2.17 (0.04) 2.18 (0.05) 2.14 (0.03) 10.47 (0.044) 10.35 (0.077) 10.28 (0.066) 10.47 (0.127) 10.18 (0.121) 3.93 (0.018) 3.93 (0.043) 3.99 (0.028) 4.02 (0.029) 4.04 (0.030) 0.009 (0.20)* 0.355 (0.34)* 0.466 (0.28)* 0.277 (0.48)* 0.912 (0.69)* Slope (SE) 0.159 (0.004) 0.151 (0.007) 0.162 (0.001) 0.172 (0.013) 0.120 (0.011) 0.041 (0.002) 0.037 (0.005) 0.033 (0.004) 0.046 (0.005) 0.027 (0.005) 0.113 (0.004) 0.113 (0.007) 0.123 (0.008) 0.122 (0.013) 0.084 (0.009) 0.006 (0.002) 0.002 (0.002)* 0.003 (0.003)* 0.004 (0.004)* 0.005 (0.005)* 0.240 (0.016) 0.161 (0.021) 0.176 (0.026) 0.169 (0.034) 0.006 (0.051)* 0).

Anterior chamber depth (mm)

Posterior segment length (mm)

TABLE 1. Gestational Age and Birth Weight for Each Stage of ROP Gestation at Birth Stage 0 1 2 3 3 Overall n 67 19 26 12 12 136 Mean (SD) 29.4 (1.87) 28.1 (1.63) 26.7 (1.73) 27.0 (2.30) 26.4 (2.05) 28.2 (2.23) n 66 19 25 12 12 134 Birth Weight Mean (SD) 1256.9 (334.99) 1042.4 (270.82) 903.1 (230.07) 944.3 (297.16) 769.3 (170.89) 1088.8 (340.37)

Lens thickness (mm)

Spherical equivalent (D)

* Nonsignicant parameter change (i.e., slope or intercept SE, standard error.

IOVS, December 2008, Vol. 49, No. 12

TABLE 4. Corneal Curvature at Term (Intercept) According to Stage of ROP Stage of ROP 0 1 2 3 3 n 67 19 26 12 12 Intercept (SE) (mm) 6.87 (0.03) 6.55 (0.06) 6.54 (0.06) 6.75 (0.11) 6.58 (0.18) Slope (SE) 0.095 (0.004) 0.079 (0.006)* 0.070 (0.005)* 0.072 (0.011)* 0.060 (0.009)* Slope 2 (SE)

Infant Eye Growth and Refractive Error Development

5201

quadratic term is included in the model, the coefcient is indicated by how much the rate of change is changing. To nd the predicted value of a parameter y at a given postmenstrual age x, given the intercept a, the slope coefcient b, and quadratic term c y a b x 40 c x 40
2

0.0034 (0.0005)

* Nonsignicant parameter change (i.e., slope or intercept SE, standard error.

0).

For a linear relationship, the last part of the equation is excluded. All statistical analyses were performed with commercial software (SPSS ver. 12 for Windows; SPSS, Chicago, IL, and SAS v. 8.1; SAS, for SunOS; Sun Microsystems).

Appropriate models were thus found for the biometric parameters and refractive state. Next, each model was retted with sex as the variable, to identify whether, at this early stage of ocular development, there was any effect of sex on ocular growth. In addition, gestation at birth and birth weight were included in the models to investigate whether differences between stages of ROP were still present when these confounders were included. Additional analyses explored the following relationships between AL, stage of ROP, and refractive error; between ACD, stage of ROP, and refractive error; between LT, stage of ROP, and refractive error; and between CC, stage of ROP, and refractive error. In all cases, after adjustment for postmenstrual age, a signicance level of 5% was assumed in the models. Estimates of the intercept a (value at term) and slope coefcient b (rate of growth) of the tted relationship are presented with their standard errors for each stage of ROP (Table 3). These estimates were calculated by tting a model for each stage separately. The slope parameter in a linear model (i.e., linear growth rate) indicates the rate of change (the number of millimeters of change per week). Where a

RESULTS
A total of 136 infants were recruited. Sixty-seven did not develop ROP. Nineteen developed stage 1, 26 stage 2, and 12 stage 3, and 12 developed threshold disease and received laser treatment (stage 3 ). Birth weight and gestational age vary signicantly according to stage of ROP (F 12.3, P 0.001 and F 15.4, P 0.001), respectively. Post hoc analysis shows that for both gestation and birth weight, stage 0 is signicantly different from all other stages, with longer average gestation and higher birth weight. The other stages do not differ signicantly from each other. Summary statistics for these variables are shown in Table 1, with age ranges at each examination and by stage of ROP in Table 2, model parameter estimates in Tables 3 and 4, and summary data of the parameters presented in Table 5. Of the total 136 infants, 99 had complete data sets collected at 32 weeks, 111 at 36 weeks, 116 at 40 weeks, 112 at 44 weeks, and 91 at 52 weeks. In addition, 46 had complete data sets for all ve time points, 43 were complete for four, 36 for

TABLE 5. Summary Statistics, at Each Time Point, for Each Stage of ROP Parameter AL (mm) Stage 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 0 1 2 3 3 T1 (32 wk) 15.4 (0.42) 15.2 (0.33) 15.1 (0.50) 15.3 (0.51) 14.9 (0.38) 2.0 (0.19) 2.1 (0.38) 2.0 (0.19) 1.9 (0.29) 2.0 (0.05) 9.6 (0.43) 9.3 (0.54) 9.3 (0.39) 9.9 (1.29) 9.1 (0.33) 3.8 (0.22) 3.8 (0.23) 3.9 (0.33) 4.0 (0.14) 3.9 (0.00) 2.1 (2.27) 2.1 (2.19) 2.0 (2.72) 1.6 (2.01) 2.9 (4.27) 6.1 (0.41) 5.9 (0.34) 5.9 (0.30) 5.7 (0.23) 4.9 (0.00) T2 (36 wk) 16.1 (0.45) 16.1 (0.37) 15.8 (0.44) 16.2 (0.63) 15.9 (0.50) 2.1 (0.32) 2.1 (0.24) 2.0 (0.21) 2.0 (0.38) 2.0 (0.17) 10.1 (0.46) 10.0 (0.28) 9.8 (0.33) 10.1 (0.40) 9.8 (0.43) 3.9 (0.18) 4.0 (0.15) 4.0 (0.21) 4.1 (0.14) 4.0 (0.11) 1.2 (2.17) 1.2 (2.15) 1.4 (1.98) 1.5 (2.22) 0.12 (3.62) 6.4 (0.24) 6.2 (0.26) 6.3 (0.28) 6.5 (0.48) 6.2 (0.18) T3 (40 wk) 16.8 (0.46) 16.6 (0.61) 16.7 (0.51) 16.8 (0.44) 16.4 (0.40) 2.3 (0.19) 2.2 (0.23) 2.2 (0.20) 2.2 (0.28) 2.1 (0.11) 10.6 (0.45) 10.5 (0.43) 10.4 (0.38) 10.6 (0.44) 10.2 (0.43) 4.0 (0.19) 4.0 (0.27) 4.0 (0.14) 4.0 (0.29) 4.1 (0.11) 0.74 (1.83) 0.07 (1.67) 0.03 (1.74) 0.55 (1.71) 1.6 (2.25) 6.9 (0.24) 6.8 (0.23) 6.6 (0.39) 6.9 (0.44) 6.7 (0.44) T4 (44 wk) 17.4 (0.48) 17.2 (0.55) 17.2 (0.57) 17.4 (0.56) 16.9 (0.65) 2.4 (0.23) 2.3 (0.22) 2.3 (0.34) 2.4 (0.18) 2.3 (0.25) 11.0 (0.45) 10.9 (0.51) 10.8 (0.48) 11.0 (0.64) 10.6 (0.53) 4.0 (0.22) 4.0 (0.15) 4.1 (0.26) 4.0 (0.19) 4.1 (0.21) 1.9 (1.76) 0.93 (1.22) 0.60 (2.18) 0.80 (2.34) 0.90 (1.56) 7.2 (0.28) 7.1 (0.29) 7.0 (0.48) 7.2 (0.24) 7.1 (0.34) T5 (52 wk) 18.6 (0.54) 18.6 (0.39) 18.6 (0.75) 18.8 (0.82) 18.2 (1.1) 2.8 (0.25) 2.8 (0.25) 2.7 (0.35) 2.8 (0.21) 2.5 (0.29) 11.8 (0.54) 11.9 (0.42) 11.9 (0.58) 12.1 (0.93) 11.4 (0.86) 4.0 (0.21) 3.9 (0.19) 4.0 (0.14) 3.9 (0.17) 4.1 (0.31) 2.1 (1.25) 1.4 (1.10) 1.7 (1.52) 1.3 (2.03) 0.65 (2.60) 7.6 (0.31) 7.5 (0.36) 7.5 (0.17) 7.5 (0.26) 7.3 (0.31)

ACD (mm)

PSL (mm)

LT (mm)

Refraction (D)

CC (mm)

Data are expressed as the mean (SD).

5202

Cook et al.

IOVS, December 2008, Vol. 49, No. 12

TABLE 6. Minimum and Maximum Number of Observations Contributing to Each Parameter at Each Examination by Stage Stage Biometric parameters and refraction (n) 0 1 2 3 3 0 1 2 3 3 T1 MinMax 5458 1415 2122 78 23 33 14 13 4 1 T2 MinMax 5253 1618 24 1011 7 44 11 12 7 3 T3 MinMax 5455 16 24 13 12 50 11 15 11 7 T4 MinMax 5354 14 22 14 12 47 6 13 10 8 T5 MinMax 38 14 13 1213 1112 27 12 11 10 8

CC (n)

three, 6 for two, and 5 for one. Overall, 125 infants had complete data sets for three or more time points, illustrating the longitudinal nature of the study. Five infants died after the study period; four of these had reached a maximum stage 2 ROP, and one had reached stage 3 and had received laser treatment. The number of valid data points for each parameter at each time point is presented in Table 6. AL showed a linear growth throughout the study period. Values differed signicantly between stages of ROP (F 2.72, P 0.033). In addition, growth rates also differed signicantly between the stages (F 3.30, P 0.014). As the stage of ROP increased, the average AL decreased (Tables 3, 5; Fig. 1). The values for stage 3 did not t the general trend, perhaps because of the small sample and the higher SEM of this group. Average ACD differed little between stages of ROP (F 2.24, P 0.068). In addition, growth rates did not differ (F 2.27, P 0.068). As the stage of ROP increased, however, the average ACD decreased, and the slope (i.e., the growth rate) also decreased. Stage 3 infants had the shallowest anterior chambers, which grew at the slowest rate (Tables 3, 5; Fig. 2). There was evidence that PSL differs between stages of ROP, as the more severe stages showed shorter posterior segments (F 2.73, P 0.0319). In addition, growth rates differed, slowing down as the severity of ROP worsened (F 2.73, P 0.0331; Tables 3, 5).

Average LT did not vary signicantly between stages of ROP (F 2.16, P 0.077). In addition, growth rates did not differ signicantly (F 1.32, P 0.267). There was very little evidence of any change in LT with time among all the groups (Tables 3, 5). The average CC differed signicantly between stages of ROP (F 4.64, P 0.002), but the growth rates did not differ (F 0.46, P 0.761)that is, all slopes remained parallel. Throughout the study, stage 3 eyes show consistently smaller radii of curvature than in all the other groups (Table 4, Fig. 3). A quadratic term was found to be appropriate for stage 0 eyes, but could not be tted for the higher stages, perhaps because of the smaller sample sizes. The coefcient for the quadratic effect is negative, indicating that the growth rate was slowing down. When all the biometric variables were tted against refraction, CC was found to be the most signicant contributor (F 35.48, P 0.0001). Stages 1, 2, and 3 showed similar rates of change of refractive error throughout the study, although as severity of disease increased, rates slowed down slightly. However, in stage 3 eyes, refractive error changed very little up to 3 months corrected age. This explains why they showed the least amount of hypermetropia by the end of the study (Tables 3, 5; Fig. 4). Throughout the study, the average refractive error did not differ signicantly between stages of ROP (F 1.43, P 0.229), although the rate of change did (F 7.69, P 0.0001).

FIGURE 1.

Changes in AL with age.

IOVS, December 2008, Vol. 49, No. 12

Infant Eye Growth and Refractive Error Development

5203

FIGURE 2. age.

Changes in ACD with

The average tted values for refractive error at term do not differ signicantly from 0.

Adjustments for Gestational Age and Birth Weight

As mentioned, gestational age and birth weight differ between stages of ROP, particularly in stage 0 compared with the other stages. Thus, it was thought important that the effect of gestational age and birth weight be added to the model to assess their association on the parameters of interest. When adding gestational age and birth weight the following occurs. There was no signicant difference in AL values between the different stages of ROP (F 1.97, P 0.102). However, growth rates differed between stages of ROP (F 3.3, P 0.014; Fig. 1). Both gestational age and birth weight were found to be signicant in this model. For ACD, there was no signicant difference in the average value between stages of ROP early on in the study (F 0.79, P 0.533), although there is still some suggestion that the growth rates differed between stages (F 2.4, P 0.056; Fig. 2). By 3 months, however, a signicant difference existed

between the stages of ROP (F 2.55, P 0.045; Table 7, Fig. 2). Again, both gestational age and birth weight were signicant in this model. Post hoc analysis indicates that stage 0 and stage 3 differed signicantly. Neither gestational age nor birth weight altered the model for LT. For CC, birth weight exerted a signicant effect in the model. There was no difference in the average value between stages of ROP (F 2.32, P 0.062). Similarly, there was no difference between the growth rates in the stages of ROP (Fig. 3). Birth weight had a signicant effect on refractive error development. Throughout the overall time course of the study, there was no difference in the average values between the stages of ROP (F 1.65, P 0.166), but the signicant difference in growth rates still existed (F 7.09, P 0.0001). However, by 3 months corrected age, a difference emerged between stage 0 and treated (3 ) eyes (F 2.16, P 0.008; Table 7, Fig. 4). When sex was tted to the models, there were no signicant differences between the sexes at this early stage.

FIGURE 3.

Changes in CC with age.

5204

Cook et al.

IOVS, December 2008, Vol. 49, No. 12

FIGURE 4. Changes in refractive error with age.

DISCUSSION
Simple cross-sectional studies comparing premature with fullterm infants have revealed that premature infants are more prone to development of myopia from an early age and may remain myopic later on in childhood and adolescence. This is known as myopia of prematurity,9 14 and it can continue to increase up to 2 years of age.15,16 Low birth weight and ROP have long been known to be implicated in the development of myopia, astigmatism, and anisometropia.13,14,1720 The risk of myopia at 12 months of age has been shown to double with each increasing stage of the disease, with a birth weight of less than 751 g contributing to a threefold increase in the risk of developing myopia.15 Fledelius has contributed much to our knowledge of the association between myopia and ROP and its treatment.10,12,21,22 Still, it is notoriously difcult to differentiate between the effects of disease and the effects of treatment. Eyes that develop more severe ROP are the ones more likely to need treatment, and both these factors are known to have an effect on ocular growth. Thus, treated infants have higher incidences of myopia than nontreated infants.16,23,24 The incidence of myopia ranges from 1% to 16% in eyes with stage 0 disease.16,25,26 If mild ROP is present, this incidence ranges from 17% to 50%,1517,25 increasing in some publications up to 100% in eyes with stage 3 disease.16 The most comprehensive prospective data come from the CRYO-ROP series of publications.16,17,27 In eyes randomized to no treatment, there was found to be an overall incidence of myopia of 21% at 1 year, falling to 16% at 4.5 years of age. The incidence of myopia in eyes with stage 0 disease was 10%; in eyes with spontaneously regressed ROP, 20%; and in eyes with severe ROP and sequelae, 80%.

Extensive comparisons have also been made between the two main treatment modalities. Of those eyes receiving treatment, those treated with cryotherapy show an increase in both the incidence and degree of myopia compared to laser-treated eyes.28 36 The biometric components that have been shown to contribute to this refractive error include a shallower ACD,21 increased lens power,14 increased CC,21,37,38 and a shorter overall AL than would be expected for the dioptric value of the eye.21,37 Later on, reports of increased PSL are noted.28,39 It seems that the early effect of growth restriction associated with ROP is followed later by a deregulation of ocular growth within the posterior segment. Our nding that AL displays a linear growth agrees with the ndings of others. Although Tucker et al.40 found the average growth rate to be 0.3 mm/wk over their study period, Laws et al.35 and Harayama et al.41 found more similar rates of growth (0.18 and 0.19 mm/wk, respectively) to our own (0.152 mm/ wk). In addition, average values for AL agree with other published data. Our value for stage 0 eyes at term of 16.84 mm compares well with that of Laws et al.35 (16.65 mm), Tucker et al.40 (16.6 mm), and OBrien and Clark42 (16.73 mm). The nding that AL changes at this early time period are inversely proportional to the severity of ROP concurs with others,20,35 such that the shortest ALs are seen in eyes that have been treated. When the effects of gestational age and body weight were removed, the difference in ALs in each stage of ROP was reduced. This nding is not so surprising when one considers that others have found no difference in AL between the stages of ROP, except in eyes with more severe ROP or that have

TABLE 7. Comparison of Ocular Parameters at 3 Months Post-Term by Stage of ROP Parameter AL (mm) ACD (mm) LT (mm) PSL (mm) Refraction (D) CC (mm) Stage 0 n 38 18.6 (0.54) 2.8 (0.25) 4.0 (0.21) 11.8 (0.54) 2.12 (1.25) 7.6 (0.31) Stage 1 n 14 18.6 (0.39) 2.8 (0.25) 3.9 (0.19) 11.9 (0.42) 1.43 (1.10) 7.5 (0.36) Stage 2 n 13 18.6 (0.75) 2.7 (0.35) 4.0 (0.14) 11.9 (0.58) 1.73 (1.52) 7.5 (0.17) Stage 3 n 12 18.8 (0.82) 2.8 (0.21) 3.9 (0.17) 12.1 (0.93) 1.30 (2.03) 7.5 (0.26) Stage 3 n 12 18.2 (1.08) 2.5 (0.29) 4.1 (0.31) 11.4 (0.86) 0.65 (2.60) 7.3 (0.31)

Data are expressed as the mean (SD).

IOVS, December 2008, Vol. 49, No. 12

TABLE 8. Proportion of AL for Each Variable at Each Stage Proportion of AL at Term (%) 0 ACD LT PSL 13.6 23.6 62.9 1 13.6 23.6 62.9 2 13.2 23.8 62.5 3 13.1 24.3 62.9

Infant Eye Growth and Refractive Error Development

5205

Proportion of AL at 3/12 Corrected Age (%) 3 13.0 24.7 62.2 0 15.1 21.3 63.6 1 15.1 21.1 63.9 2 14.5 21.3 64.2 3 14.9 20.8 64.3 3 3.7 2.5 2.9

undergone treatment.20,28,35 In addition, their studies measured values at times later than those in our study, when any differences may be expected to be greater. ACD changed very little relative to AL as a whole (Table 8). This may explain why ACD is not strongly associated with refractive error during this early time period. Kent et al.28 found the shallowest anterior chambers in the treated eyes, especially the cryotherapy group. Their results agree with this study, although values are not directly comparable due to differences in age at examination. CC underwent a greater degree of change during this early time period than did the other variables, and exerted more of an inuence on refractive error. The relationship between steeper CC and more severe ROP has been known for a long time.43 Our data for CC at term (6.87 mm for stage 0) compare well with those of Inagaki44 taken at 38.3 weeks (6.8 mm). Snir et al.38 measured eyes with mild regressed ROP at term as 6.84 mm (49.45 D). Their result compares with an average value of 6.65 mm in eyes with stage 2 or less in our study. Blomdahl45 measured full-term infants at 2 to 4 days after birth, and his result of 7.0 mm highlights the difference between premature and full-term eyes. When each variable is studied as a proportion of the total AL, as time goes by, some interesting trends emerge (Table 8). ACD increases as a proportion of AL throughout the study period. However, the tendency for it to do so is much less in the eyes that have been treated with the laser (13%13.7% in stage 3 eyes, compared with 13.6%15.1% in stage 0 eyes). Refractive error changes show that by 3 months postmenstrual age, stage 3 eyes have less hypermetropia than at other stages, with the difference being greater in the treated eyes. The nding that only insignicant differences exist between the milder stages has been reported previously.20,26 At 3 months, treated eyes showed an average value of 0.648 D, which is less myopic than Laws and Clark33 found at the same time ( 3.25 D). However, two points should be borne in mind. First, there is a time difference between these two studies of 3 to 4 years. During this time, there have been advances in neonatal care and outcomes. This may well have an inuence on the growth of the eye at this early time. Second, one of the infants (infant A) who received laser treatment in this study group showed consistently high levels of hypermetropia throughout the study period. When this case
TABLE 9. Infant A Data Compared with the Stage 3 Variable Gestational age (wk) Birth weight (g) AL (mm) ACD (mm) LT (mm) CC (mm) Infant A 25 680 17.13 2.46 4.07 7.2 Group Average Eye

was analyzed in more detail, interesting differences were revealed (Table 9). It is apparent that this infant was born relatively early, with very low birth weight. The AL was signicantly shorter than average for the group of treated eyes. More specically, the anterior chamber was very much shallower than that of the rest of the group. CC and LT did not differ signicantly. The former differences could quite conceivably account for the refractive error discrepancy. If the value for this infant is removed from the data, the average value for refractive error is less hypermetropic ( 0.24 D) for treated eyes at 3/12 corrected age. The refractive status of this cohort of infants can be compared with those of full-term infants at similar ages (Table 10). There is consistently less hypermetropia noted at term in premature infants, with or without ROP, when compared with full-term infants. By 3 months corrected age, this difference is much less in those premature eyes without ROP. Eyes with ROP, however, still maintain less hypermetropia than full-term eyes by the end of the study period. Figure 5 shows the frequency distribution of refractive error for the subgroup of eyes without ROP from this study at 3 months corrected age. These data have been compared with data from two other studies. The data from Mayer et al.46 consist of data from full-term eyes. The data from Quinn et al.17 include only those eyes without ROP. The graph highlights the similarity of the data sets at 3 months; conrming the attempt at emmetropization by premature eyes without ROP. The exact cause of the alteration in refractive error development associated with ROP is not fully understood. In 1955, Birge13 recognized that children who had been born prematurely developed a more severe form of myopia than the usual developmental type of refractive error. He referenced the earlier work by Szewczyk18 and Ryan in 1952,47 who looked at the role of oxygen in premature birth, and postulated that the sudden withdrawal of oxygen precipitated a vascular crisis that affected the vascularity of the choroid, resulting in myopia. Fielder48 later postulated that the ROP lesion, being located in the part of the eye undergoing maximum growth during late fetal and early neonatal life, may exert a mechanical effect on the anterior sclera and anterior segment. This would t with
TABLE 10. Comparison of Refractive Errors in Premature and Full-Term Eyes at Term and 3 Months Study This study Blomdahl45 Wood et al.7 Mayer et al.51 Kent et al.28 SE, spherical equivalent. Stage 1 Stage 3 SE at Term (D) 0 0.9 3.6 0.5 2.2 SE at 3/12 (D) 2.12 0.65 2.44 2.75

Average Stage 3 26 769 18.16 2.53 4.09 7.3

Infant born very early with very low birth weight. Showed consistent hypermetropia compared to rest of group 3 . The anterior chamber was very shallow.

5206

Cook et al.

IOVS, December 2008, Vol. 49, No. 12

FIGURE 5. Frequency distribution of refractive errors at 3 months corrected age in premature eyes without ROP in the present study, compared with that in other studies.

the fact that the axis of astigmatism is seen to rotate as ROP severity increases,20 and with the theory of anterior segment growth arrest, of which the hallmark is a shallower anterior chamber, and more highly curved cornea. It is clear that the present study supports the theory of anterior segment growth arrest, as eyes with more severe ROP developed shallower anterior chambers with more highly curved corneas. Gestational age and birth weight clearly both have an early effect on ocular growth, but they do not explain all the differences in the growth rates of the biometric variables and refractive error demonstrated by this study. Retinopathy of prematurity itself appears exert an inuence on the growth of the eye at this early stage of development. Also signicant during this early time period is the effect that treatment for ROP may be starting to demonstrate. It is important to consider, however, that there may be other environmental factors still unaccounted for that affect ocular development at such a sensitive time.

References
1. Pararajasegaram R. Vision 2020 : the right to sight: from strategies to action. Am J Ophthalmol. 1999;182:359 360. 2. Zadnik K, Satariano WA, Mutti DO, Sholtz RI, Adams AJ. The effect of parental history of myopia on childrens eye size. JAMA. 1994; 271l:13231332. 3. Lee KE, Klein BE, Klein R, Fine JP. Aggregation of refractive error and 5 year changes in refractive error among families in the Beaver Dam Eye Study. Arch Ophthalmol. 2001;119(11):1679 1685. 4. Wu SY, Nemesure B, Leske MC. Refractive errors in a black adult population: the Barbados Eye Study. Invest Ophthalmol Vis Sci. 1999;40(10):2179 2184. 5. Cook A, White S, Batterbury M, Clark D. Ocular growth and refractive error development in premature infants without ROP. Invest Ophthalmol Vis Sci. 2003;44:953960. 6. Butcher J, OBrien, C. The reproducibility of biometry and keratometry measurements. Eye. 1991;5:708 711. 7. Wood I, Mutti D, Zadnik K. Crystalline lens parameters in infancy. Ophthalmol Physiol Opt. 1996;16(4):310 317. 8. The Committee for the Classication of ROP. An international classication of ROP. Arch Ophthalmol. 1984;102:1130 1134. 9. OConnor A, Stephenson T, Johnson A, et al. Long term ophthalmic outcome of low birth weight children with and without retinopathy of prematurity. Paediatrics. 2002;109(1):1218. 10. Fledelius HC. Myopia of prematurity: changes during adolescence. Doc Ophthalmol Proc Series.1981;28:63 69.

11. Hungerford J, Stewart A, Hope P. Ocular sequelae of preterm birth and their relation to ultrasound evidence of cerebral damage. Br J Ophthalmol. 1986;70:463 468. 12. Fledelius HC. Pre-term delivery and subsequent ocular development: a 710 year follow up of children screened for ROP 1982 4. Acta Ophthalmol Scand. 1996;74(3):297300. 13. Birge H. Myopia caused by prematurity. Trans Am Ophthalmol Soc 1955;292298. 14. Gordon A, Donzis P. Myopia associated with ROP. Ophthalmology. 1986;93:15931598. 15. Page J, Schneeweiss S, Whyte H, Harvey P. Ocular sequelae in premature infants. Paediatrics. 1993;92(6):787790. 16. Quinn G, Dobson V, Repka MX, et al. The CROP-ROP Group. Development of myopia in infants with birthweights less than 1251 g. Ophthalmology. 1992;99(3):329 340. 17. Quinn G, Dobson V, Kivlin J et al. The Cryo-ROP Group. Prevalence of myopia between 3 months and 5 12 years in premature infants with and without ROP. Ophthalmology. 1998;105(7):12921300. 18. Szewczyk TS. Retrolental broplasia and related ocular diseases. Am J Ophthalmol. 1953;36:1336 1361. 19. Robinson R, OKeefe M. Follow-up study on premature infants with and without ROP. Br J Ophthalmol. 1993;7:9194. 20. Laws D, Shaw DE, Robinson J, Jones HS, Ng YK, Fielder AR. Retinopathy of prematurity: a prospective study: review at 6 months. Eye. 1992;6:477 483. 21. Fledelius HC. Pre-term delivery and subsequent ocular development: 710 year follow-up of children screened 1982 84 for ROP. 4. Oculometric and other metric considerations. Acta Ophthalmol Scand. 1996;74(3):301305. 22. Fledelius HC. Ophthalmic changes from 10 18 years: a long -term study of sequels to low birth weight refraction. Acta Ophthalmol (Copenh). 1980;58(6):889 898. 23. Holmstrom,G, Azazi M, Kugelberg U. Ophthalmic long term follow-up of preterm infants: a population based, prospective study of refraction and its development Br J Ophthalmol. 1998:82:1265 1271. 24. Nissenkorn I, Yassur Y, Mashkowski D, Sherf I, Ben-Sira I. Myopia in premature babies with and without ROP. Br J Ophthalmol. 1983;67:170 173. 25. Schaffer D, Quinn G, Johnson L. Sequelae of arrested mild ROP. Arch Ophthalmol. 1984;102:373376. 26. Darlow B, Clemett RS, Horwood LJ, Mogridge N. Prospective study of New Zealand infants with birth weight less than 1500g and screened fro ROP: visual outcome at age 7 8 years. Br J Ophthalmol. 1997;81:935940. 27. Quinn GE, Dobson V, Siatkowski R, et al. Does Cryotherapy affect refractive error?results from treated versus control eyes in the

IOVS, December 2008, Vol. 49, No. 12

Infant Eye Growth and Refractive Error Development

5207

28. 29.

30.

31.

32.

33. 34. 35.

36.

37.

cryotherapy for retinopathy of prematurity trial. Ophthalmology. 2001;108:343347. Kent D, Pennie F, Laws D, White S, Clark D. The inuence of ROP on ocular growth. Eye. 2000;14:2329. Connolly B, McNamara JA, Sharma S, Regillo CD, Tasman W. A comparison of laser photocoagulation with trans-scleral cryotherapy in the treatment of threshold ROP. Ophthalmology. 1998; 105(9):1628 1631. White J, Repka M. Randomised comparison of diode laser photocoagulation versus cryotherapy for threshold ROP: 3 year outcome. J Paediatr Ophthalmol Strabismus. 1997;43:83 87. Algawi K, Goggin M, OKeefe M. Refractive outcome following diode versus cryotherapy for eyes with ROP. Br J Ophthalmol. 1994;78:612 614. Knight-Nanan D, OKeefe M. Refractive outcome in eyes with ROP treated with cryotherapy or diode laser: 3 year follow up. Br J Ophthalmol. 1996;80:998 1001. Laws D, Clark D. Cryotherapy and laser for ROP: refractive error outcomes; a longitudinal study. Br J Ophthalmol. 1997;81:1215. OKeefe M. Refractive outcome in eyes with ROP with cryo or diode. Br J Ophthalmol. 1996;80:998 1001. Laws DE, Haslett R, Ashby D, OBrien C, Clark D. Axial length biometry in infants with retinopathy of prematurity. Eye. 1994;8: 427 430. Shalev B, Farr A, Repka M. Randomised comparison of diode laser photocoagulation versus cryotherapy for threshold ROP: seven year outcome. Am J Ophthalmol. 2001;132:76 80. Gallo J, Fagerholm P. Low grade myopia in children with regressed ROP Acta Ophthalmol. 1993;71:519 523.

38. Snir M, Friling R, Weinberger D, Sherf I, Axer-Siegel R. Refraction and keratometry in 40 week old premature and term infants. Br J Ophthalmol. 2004;88:900 904. 39. Fledelius HC. Ophthalmic changes from 10 18 years: a longitudinal study of sequels to low birth weight. 4. Ultrasound oculometry of vitreous and axial length. Acta Ophthalmol (Copenh). 1982; 60(3):403 411. 40. Tucker SM, Enzenauer RW, Levin A, Morin JD, Hellman J. Corneal diameter, AL, and IOP in premature infants. Ophthalmology. 1992; 99(8):1296 1300. 41. Harayama, K, Amemiya, T, Nishimura, H. Development of the eyeball during foetal life. J Paediatr Ophthalmol Strabismus. 1981;18:37 40. 42. OBrien C, Clark D. Ocular biometry in pre-term infants without ROP. Eye. 1994;8:662 665. 43. Hittner H, Rhodes L, McPherson A. Anterior segment abnormalities in cicatricial ROP. Ophthalmology. 1979;86:803 816. 44. Inagaki Y. The rapid change in corneal curvature in the neonatal period and infancy. Arch Ophthalmol. 1986;104:1026 1027. 45. Blomdahl S. Ultra-sound measurement of the eye in the new-born infant Acta Ophthalmol. 1979;31:1048 1056. 46. Mayer DL, Hansen RM, Moore BD, Kim S, Fulton AB. Cycloplegic refractions in healthy children ages 1 through 48 months. Arch Ophthalmol. 2001;119:16251628. 47. Ryan H. Retrolental broplasia: a clinicopathological study. Am J Ophthalmol. 1952;35:329 341. 48. Fielder A, Quinn G. Myopia of prematurity: Nature, nurture or disease? Br J Ophthalmol. 1997;81:23.