Long-Term Ophthalmic Outcome of Low Birth Weight Children With and Without Retinopathy of Prematurity

METHODS

RESULTS

From the original cohort of 505 low birth weight infants, 29 died after the original study was completed, 5 had moved outside the United Kingdom, and 23 could not be traced. At the GP’s request, the families of 2 children were not contacted. Among the remaining 446 children, there were 16 parental refusals and 176 nonresponders despite repeated reminders, leaving 254 who consented.

There were no statistically significant differences in mean gestational age, birth weight distribution, and incidence of ROP between the 254 tested and the 222 children not tested but known to be alive (Table 1). Table 1 shows the distribution of the stages of ROP but does not include cicatricial ROP as this cohort is biased toward mild ROP; only 5 children in the original study and 2 of the follow-up study showed cicatricial changes.

There was no significant difference between the study and control (n = 169) groups for social class, but the mean age of the study cohort (11.51 years [standard deviation (SD) 0.56]) was slightly higher than for the control subjects (11.36 years [SD 0.32]; P < .05).

It was not possible to assess accurately all visual functions of 4 children of the low birth weight cohort, who had multiple disabilities including severe vision impairment. Data are available for strabismus and refractive error in these children. Three of these children had severe ROP (stage 3 or above), and the other child had stage 2 ROP.

Distance and near visual acuities and contrast sensitivity of the study cohort were statistically significantly lower than the control subjects’ for both uniocular and binocular (in all cases, P < .001, Mann-Whitney; Table 2, comparison of last 2 columns). Acuity was measured using a pinhole when reduced as this identifies any uncorrected refractive errors and therefore prevents these cases from influencing the results. However, even for this cohort of infants with birth weight of <1701 g, the median distance acuity was better than 0.0 log units (equivalent to 20/20 Snellen; by convention, 20/20 or better is considered normal). Table 3 shows the proportion of children with below-normal acuity.

Table 2 shows the effect of ROP on visual acuity and contrast sensitivity. The 4 children with multiple disabilities are not included in this table. For uniocular measures, the maximum stage of ROP in that eye was used; for binocular measures, the maximum stage of ROP in either eye was used. Children who had the most severe level of ROP (stage 3 or more) had the lowest visual acuities, whereas contrast sensitivity did not seem to be related to the stage of ROP. Left eye and binocular visual acuities for near and distance were reduced by severe ROP (stage 3/4; P < .001) compared with those with mild or no ROP. The visual acuities of the right eye were reduced by severe ROP but did not reach statistical significance. Univariate analysis of variance (by stage of ROP) showed that ROP diagnosed in the neonatal period was a poor predictor of visual function 10 to 12 years later (maximum adjusted R² = 0.078). Analysis was repeated adjusting for birth weight, but the outcome did not change, indicating that left and binocular acuities were reduced by severe but not by mild ROP (stages 1 and 2) over and above the effect of low birth weight per se.

For this study, strabismus was defined as the presence of a manifest deviation in the primary position at any distance, with or without glasses. The prevalence of strabismus in the study cohort (19.3%; n = 49) was significantly higher compared with the control group (3%; n = 5 [χ², P < .001]). The prevalence of strabismus increased with increasing ROP severity as demonstrated in Fig 1.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

The prevalence of strabismus according to stage of ROP.

The median stereoacuity (measure of stereopsis) of the low birth weight and control groups was 60 seconds of arc (normal). When grouped together by grading of stereoacuity (Table 4), analysis showed a significant difference between the 2 groups (χ², P < .001). As the stereo test used (TNO) is limited to measuring a specific range of stereoacuities, there were children outside this scale. Therefore, to include all children in the analysis, even if there was no demonstrable stereoacuity, they were coded by an arbitrary figure of 9999. After removal of all cases of strabismus from the analysis, there was no longer any difference between the low birth weight and control groups (P = .5). Stereoacuity was significantly associated with maximum stage of ROP (P = .001); median values for no ROP and stages 1 and 2 were 60 inches and 9999 for the severe ROP group. Analysis of covariance showed that this association was attributable to the reduction in the severe ROP group only.

Color vision defects were present in 0.79% (n = 2) of the low birth weight group compared with 2.4% (n = 4) of the controls. Further statistical analysis was not appropriate because of small frequencies. The only visual field abnormality detected was a homonymous hemianopia in 1 child with a hemiplegia in the low birth weight cohort.

By a combination of autorefraction and pinhole acuity, we were able to define which control children had myopia but not to quantify its magnitude precisely (pinhole overcomes but does not measure a refractive error). The study cohort demonstrated a greater prevalence of myopia (22.4%; n = 57), which we defined as all negative mean spherical equivalents (<0.0 dioptre spheres) compared with the control group (8.9%; n = 15 [χ², P < .001]). The degree of myopia in the follow-up study cohort was mild (<−3.00 diopters) in 45 children and moderate (worse than −3.00 diopters) in 12 children (classified by worst eye). The prevalence of myopia was similar in those with no or mild ROP but increased significantly in severe ROP (P = .04, χ², prevalence of myopia with no ROP = 22% [n = 28], stage 1 = 16.7% [n = 14], stage 2 = 20.5% [n = 7], stage 3/4 = 80% [n = 8]). All ocular dimensions measured differed significantly from age-control normals¹³ (published data give measurements for right eye only), and, apart from lens thickness, all were smaller. Thus, eye size of the study cohort was significantly smaller than published normative data in children of the same age (Table 5). Analysis by Kruskal-Wallis shows no significant difference in any of the ocular components within the study cohort by stage of ROP, although there is a trend of increased axial length associated with severe ROP.

To summarize the ophthalmic data for the purposes of this study, we defined ophthalmic morbidity as visual acuity below the accepted norm (0.0 log units) or the presence of strabismus, myopia, color vision defect, or visual field defect. Such an outcome occurred in 50.8% (n = 129) of the low birth weight cohort at age 10 to 12 years compared with 19.5% (n = 33) of the control group (χ², P < .001). The 4 children with unrecordable acuities were included in the ophthalmic morbidity group as they all had either myopia or strabismus. There was a significant association between ROP and ophthalmic morbidity (P = .02, χ², prevalence of ophthalmic morbidity with no ROP = 46% [n = 58), stage 1 = 48.9% [n = 41], stage 2 = 58.8% [n = 20], stage 3/4 = 100% [n = 10]).

To facilitate comparison with other studies, Table 6 shows visual impairments grouped by birth weight and ROP status. There is no significant difference in the rate of ophthalmic morbidity between birth weight categories (χ² = 4.831, P = .2). The results are not presented by gestational age categories as the original study criteria were based solely on the birth weight. Therefore, the higher gestational age groups will be incomplete and nonrepresentative.

DISCUSSION

In the mid-1980s, we undertook a study of ROP in 505 infants who weighed <1701 g at birth. Here we report the ophthalmic findings of 254 children from this cohort at age 10 to 12 years. As this follow-up was not conceived at the time of the original investigation, family contact was not maintained and this long gap without communication probably accounts for some of the response failure. At the age of 10 to 12 years, the child naturally has some say in the decision about participation, and this factor is also likely to have reduced recruitment. Accordingly, 54.6% of the children traced were assessed at 10 to 12 years. Pennefather et al¹⁴ showed that those who were not willing to participate in follow-up studies of prematurity were more likely to have problems. In contrast, Campbell et al¹⁵ showed that in their cohort of very low birth weight children, those who had lower birth weights and therefore were at greater risk of subsequent problems were more likely to attend. The study of Campbell et al, however, is a retrospective analysis of the normal clinic follow-up appointments where there is more of a perceived benefit to the child. To determine the possible effect of the low rate of response (n = 222 [46.6%]) we performed a sensitivity analysis. This showed that the rate of visual impairments in those who were not followed up would have to be as low as 3 out of 222 (1.4%) when combined with those who were followed up to overturn the statistical significance of the study cohort rate (50.8%) versus the control rate (19.5%).

Previous studies all showed that low birth weight children are at risk of developing a range of ophthalmic morbidity.^{3–5,16–37} This includes deficits of visual acuity,^26,27 contrast sensitivity^5,28 and color vision,^5,29 retarded ocular growth,^30,31 and refractive errors.^4,22,^32, Low birth weight children are also more likely to develop strabismus^33–35 and severe visual impairments as a result of either ROP³⁶ or damage to the posterior visual pathway.^33,37 Although severe ROP is widely known to reduce vision, whether visual acuity is affected by mild ROP is unknown, as in many studies there has been underascertainment of mild ROP. The value of our study is that we have been able to answer this question. In our study, children with no or regressed mild ROP had similar corrected acuities. These acuities were very slightly but significantly reduced (P < .001) compared with the control group. However, this difference is small (mean difference and 95% confidence interval for binocular distance acuity = 0.06 [0.03, 0.07]), and the mean acuity was better than 0.0 log units (normal). Therefore, no or regressed mild ROP seems to have no important long-term effect on visual acuity. The only pertinent comparison is with the CRYO-ROP Study,³⁸ which reported that visual acuity at age 4.5 years was not significantly affected by prethreshold ROP (intracohort comparison). However, this group did not include a comparison with controls and at this young age might not have identified the subtle effects observed by us at age 10 to 12 years.

Myopia is a widely known consequence of low birth weight, especially after severe ROP. Almost 2 decades ago³⁹ and again more recently,⁴⁰ Fledelius also observed that compared with children who were born at term, children who were born prematurely and had little or no ROP were also more likely to have myopia. The uncomfortable term “myopia of prematurity” was coined to describe this latter group and to differentiate it from both myopia associated with severe ROP and also myopia in individuals who were not born preterm, ie, the most frequently encountered type of myopia in the population. Myopia of prematurity has the following characteristics: a relatively highly curved cornea, shallow anterior chamber, and thick lens with an axial length that is shorter than would be expected for the dioptric value. We confirm here that compared with term controls, children who were born prematurely but had no or only mild ROP do have altered growth of the eye, and this contributes to the high incidence of myopia in this population. Thus, these are not just small eyes in small children; this morphology is suggestive of arrested rather than subsequently disturbed ocular growth,^31,41 and this perturbation affects mainly the anterior segment of the eye. This is in contrast to myopia not associated with prematurity, which is attributable to increased axial length.

Contrast sensitivity was significantly reduced in our study cohort compared with control subjects, but there was no additional effect associated with ROP. The Pelli-Robson test uses letters of low spatial frequency (large letters, all 20/900), and its reduction in children with normal-to-mild visual acuity deficits therefore reflects not retinal but central neurologic function.⁴² Although this effect is small, it may signify a subtle underlying adverse effect of preterm birth or postnatal events on central neurologic development.

The greatest increase in visual impairments is the 6-fold difference in the prevalence of strabismus in the cohort compared with the control group. The presence of strabismus is associated with increasing severity of ROP. However, the retinopathy may also be associated with central neurologic damage, and here it is not possible to differentiate fully these two mechanisms. This will be analyzed in detail in a future publication.

By including relatively large infants of birth weight of <1701 g, our cohort contained a preponderance of infants with no or mild ROP and was suitable therefore to differentiate the effects on the visual system of low birth weight alone and ROP. This contrasts with and complements those studies of severe ROP, most notably the US Multicenter Study for Cryotherapy for Retinopathy of Prematurity.^{7,11,29,32,36,37} Our findings confirm that low birth weight children have a number of long-term ophthalmic sequelae. These include altered eye growth, reduced visual functions, and an increased risk of strabismus.

This study demonstrates that preterm birth alone has an impact on the immature visual system from the eye to the cortex. This could be attributable to the removal of the fetus from a uterine milieu uniquely suited to promote growth and the subsequent exposure of immature tissues to an environment that is quite unlike that experienced at any other period of life with respect to environment and stimulation.⁴³ These effects could be mediated both centrally in the central nervous system (eg, altered contrast sensitivity and strabismus) and peripherally in the eye (eg, altered eye growth, reduced visual acuity).

More than half of this low birth weight cohort at 10 to 12 years had an ophthalmic problem, including strabismus and myopia, which would in most cases result in a consultation with an optometrist or referral to an ophthalmologist. Whether the referral is to the hospital or to the community ophthalmic services, the number of referrals significantly impacts on health care providers. Although severe ROP is associated with the highest rate of visual impairments, this study demonstrates that many sequelae are attributable to prematurity per se, and mild ROP confers no additional risk for any of these adverse effects apart from strabismus. Not enough is known about the functional significance of the ophthalmic sequelae and whether they create an obstacle to learning and school attainment. The children in this low birth weight cohort are currently undergoing psychometric assessment to determine this. Fortunately, apart from 4 multiply handicapped visually impaired children, it is reassuring that in this cohort there were relatively few severe untreatable visual impairments.