|Home | About | Journals | Submit | Contact Us | Français|
Few studies of low birthweight children have explored the relationship between later visual morbidity and neuropsychological function. This study evaluated these outcomes using a geographically defined cohort.
Prospective study of retinopathy of prematurity (ROP) in infants born weighing <1701 g, undertaken in 1985–7. 254 of the survivors consented to ophthalmic examination at 10–13 years. Four children were severely disabled and could not complete the tests. 198 of the remaining agreed to neuropsychological assessment at 11–14 years (British Ability Scales II (BAS), Movement Assessment Battery (ABC), Neale Analysis of Reading Ability).
At 10–13 years, 99/198 children had an adverse ophthalmic outcome (AOO) (reduced acuity n=48, myopia n=40, strabismus n=36, colour defect n=2, field defect n=1). There were no significant differences between children with AOO and those with a normal ophthalmic outcome with regard to sex, gestation, birth weight, neonatal cranial scan appearances and social class. 106/198 had ROP; 98 had mild ROP with no increased risk of AOO in later childhood. All eight children with severe ROP had an AOO in later childhood. Children with an AOO performed worse on the BAS, ABC and reading ability tests.
At age 10–13, 50% of children born <1701 g have an AOO. These children are not simply those with earlier gestations, lower birth weight or ROP. Children with AOO have a worse neuropsychological outcome. The next step is to determine whether there are visual interventions which can improve ophthalmic outcome and whether a better neuropsychological outcome follows.
Survivors of preterm birth are at increased risk of cognitive impairment1,2,3,4,5,6,7,8,9 and poor school performance.2,3,7,9,10 In particular, specific difficulties in reading, spelling, number skills and handwriting have been described. Low birthweight children are also at risk of later vision disorders, such as strabismus, defective visual acuity and myopia,11,12,13,14,15,16,17,18 whether or not they had retinopathy of prematurity (ROP).19 The relationship between disorders of vision and cognitive deficits is not well understood as the two areas of function have rarely been studied in detail and concurrently in the same children.20 In low birthweight children, it is not known if vision and cognitive disorders have a common causal pathway or whether problems in the visual pathways impair specific areas of school performance.
We studied this relationship in a geographically defined cohort of low birthweight children to test the hypothesis that in school‐aged children who were born with low birth weight, there is an association between neuropsychological outcome and visual morbidity, and that this association is independent of social and perinatal factors.
A geographically defined, prospective study of ROP was undertaken in 1985–7.21 Infants born to mothers resident in the East Midlands region in the UK, with birth weight <1701 g and surviving three weeks, were enrolled from five neonatal units in Nottingham, Derby and Leicester. This cohort of 505 infants (mean gestation 31.1 weeks; mean birth weight 1345 g) had ophthalmic examinations from 3 to 12 weeks of age and at 6 months corrected age. Of these, 248 developed ROP, of whom 4% had severe ROP (stage 3 or worse).
Of the 505 infants, 29 had died, 5 had left the UK and 23 could not be traced. Two families were not contacted at their doctor's request. Sixteen families declined to participate and 176 did not give written consent despite repeated reminders (three postal and two telephone). Thus 254 families consented to ophthalmic examination, conducted by one orthoptist.19 However, four children were too disabled to undertake neuropsychological assessments. Of 250 families, 198 gave written informed consent to neuropsychological assessment (approximately one year after ophthalmic assessment) conducted by one psychologist. All investigators were masked to the child's perinatal history and original ROP status. We chose not to study a matched control group of term infants as we were principally interested in within‐cohort analyses.
The ophthalmic tests are described in detail elsewhere.19 In summary, these were: visual functions (logMAR distance and near acuity,22,23 contrast sensitivity, stereoacuity, perimetry, colour vision); strabismus (cover test and prism tests); and refractive state and eye dimensions. Corrected visual acuity was assessed using the child's own spectacles or a pinhole if refractive error had not been previously detected or acuity remained below 0.0 logMAR units even with glasses.
The overall measure of the British Ability Scales II (BAS II)24 is the general conceptual ability (GCA) score with mean (SD) of 100 (15) in the normal population. The GCA score sums three clusters: verbal, non‐verbal and spatial abilities. Individual BAS items have a mean of 50 (10). In addition, there are “diagnostic scales” which do not contribute towards the GCA score (eg speed of information processing).
The Movement Assessment Battery for Children (ABC) 25 tests dexterity, ball skills and balance. A score below the 5th percentile of the normal population is classified as a definite motor problem, and between 5th and 15th percentiles as borderline.
The Neale Analysis of Reading Ability—Revised is an oral test for children aged 6–12 years to assess accuracy, rate and comprehension.26
We calculated either the means and standard deviations or, for variables not normally distributed, medians and interquartile ranges (IQRs). All data were investigated for normality, by histograms, to determine whether parametric or non‐parametric methods should be used. The unpaired t test was used for normally distributed variables and the Mann–Whitney U test for variables not normally distributed. χ2 tests were performed on categorical data. When comparing more than two groups, one‐way analysis of variance or the Kruskal–Wallis test was used for continuous data and the χ2 test for categorical data. Multivariate linear regression was used to control for potential confounders. Testing was in all cases two‐sided. We analysed all data using the statistics packages SPSS (version 8) and STATA (version 5).
Ophthalmic assessment was carried out between the ages of 10 and 13 years (mean 11 years, 6 months). Neuropsychological assessment was carried out between the ages of 11 and 14 years (mean 12 years, 8 months). Of the 198 children who had a neuropsychological assessment, 174 were white, 12 were Asian, 6 were black and 6 were mixed race. We did not find any significant differences between the 198 children assessed and the 307 not assessed with regard to birth weight, gestational age, sex, incidence of ROP or cranial ultrasound abnormalities (table 11).). Parental occupation was not recorded at birth. Current social class of children not assessed was not derived because their parents declined consent to providing information.
Of the 198 children, 99 (50%) were classified as having an adverse ophthalmic outcome at 10–13 years.19 An adverse ophthalmic outcome was defined as corrected binocular visual acuity below the norm (6/6 Snellen, 20/20 in the USA or 0.0 logMAR units) (n=48, 24%) or the presence of strabismus (n= 36, 18%), myopia (n=40, 20%), colour vision defect (n=2, 1%) or field defect (n=1, 0.5%). Some children had more than one defect. These outcomes result in a referral to ophthalmic services in the UK. All children who had strabismus before 7 months corrected age still had strabismus at 12 years. Even those who had surgery still had some residual deviation. We found no statistically significant differences between the children with an adverse ophthalmic outcome and those with a normal ophthalmic outcome with regard to sex, gestational age, birth weight, neonatal cranial ultrasound scan appearances and current social class (table 22).
Of the 198 children, 106 had ROP in the neonatal period. Unsurprisingly, children with ROP were smaller and less mature at birth (table 33).). The presence or stage of ROP in the neonatal period, excluding the eight children with severe ROP, did not predict which children would have an adverse ophthalmic outcome at 10–13 years (table 22).). On excluding these eight with stage 3/4 ROP, there was no longer a significant association between ROP status and adverse ophthalmic outcome (p=0.45 rather than 0.013).
First, the cohort performed below average in almost all areas (table 44).). However, children who had ROP in the neonatal period did not have lower BAS II scores overall 11–14 years later.
Second, we compared BAS II scores for children with and without an adverse ophthalmic outcome. There was an association between long‐term ophthalmic outcome and performance on the BAS II in almost all areas (statistically significant for spatial ability, p=0.01). Excluding the eight with stage 3/4 ROP (all had an adverse ophthalmic outcome), this association between adverse ophthalmic outcome and spatial ability remained strong (p=0.02). Children with an adverse ophthalmic outcome performed significantly worse on both design recall and pattern construction within the spatial ability cluster (p=0.037 and p=0.009, respectively). Those with an adverse ophthalmic outcome performed worse on speed of information processing and in recall within the diagnostic scales.
Third, the results of multivariate linear regression showed that after controlling for ROP stage, gestational age, birth weight, cranial ultrasound, social class and sex, the BAS scores (table 55)) were still associated with adverse ophthalmic outcome. The associations were, therefore, independent of these potential confounding factors, insofar as we were able to test them. A cerebral scan was not done for 39 (39.9%) of the children with normal ophthalmic outcome and 40 (40.4%) with adverse ophthalmic outcome. (Scanning was not routinely done in the East Midlands in 1985–7.) Maternal social class data were not available for 24 (24.2%) of the children with normal ophthalmic outcome and 18 (18.2%) with adverse ophthalmic outcome because the parent declined this information. Oxygen dependence at 28 or 36 weeks was not recorded in the original dataset.
Fourth, the presence of strabismus, abnormal three‐dimensional vision (stereopsis) and decreased visual acuity with correction were all significantly associated with poor spatial ability (see table 66 for the p vaues). Binocular contrast sensitivity with correction and refractive error were not associated with BAS II spatial ability scores (data not shown).
Fifth, neonatal ROP status was examined in relation to BAS II performance (table 77).). Only 8 (4%) of the cohort had severe ROP (stage 3 or above), none had stage 5 ROP (total retinal detachment), and none were registered blind. Those children with ROP stage 3 or 4 achieved lower BAS II GCA and cluster scores. There was no systematic trend in scores between those who never had ROP and those with stages 1 and 2.
Children with an adverse ophthalmic outcome performed less well in all three areas of reading ability than those with normal ophthalmic outcome (up to 0.3 SD worse), but this difference was not statistically significant (data not shown).
In movement ability tests, 113 (57%) of the cohort scored below the 15th percentile for the normal population (data not shown). Children with an adverse ophthalmic outcome had worse median scores in all four areas of the ABC (statistically significant for ball skills, p=0.01).
Our geographical cohort study is unusual because it combines neonatal and long‐term ophthalmic data with later neuropsychological data. The return rate of 254 children examined at 11–14 years out of 505 at 3–12 weeks is a criticism, but it follows 10 years of no contact. Other follow‐up studies from preterm to teenage show comparable rates.28 There were no statistically significant differences in gestational age, birth weight and incidence of ROP between the 254 tested and the 222 children not tested but alive.19 We suggest therefore that follow‐up of these 254 children over a long period is both valuable and representative of the original cohort. Sensitivity analysis showed that the adverse ophthalmic outcome rate in those not followed up would have to be 3/222 (1.4%) to overturn the statistical significance of the study cohort rate of adverse ophthalmic outcome (50%).
An adverse ophthalmic outcome at age 10–13 years was found in 99/198 children examined (50%); 18% had strabismus, consistent with other studies.29,30 Adverse ophthalmic outcome was not simply a product of being born more preterm or being born more socially deprived (table 22).). This confirms the findings of Powls et al17 but they were unable to exclude the possible confounder of neonatal ROP as their very low birthweight cohort were not screened ophthalmologically. In our study, 106 children had ROP in the neonatal period (table 33).). ROP still occurs,31 but stage 1 and 2 ROP do not predict ophthalmic problems 10–13 years later (table 22).
Only one other long‐term, follow‐up, geographical study of ophthalmic and cognitive outcomes in a cohort of preterm infants has been published.20 In that study, 279 children out of a potential 382 born before 32 weeks' gestation in Liverpool during 1991–2 and attending mainstream schools (thus excluding children with major disability) were re‐examined at seven years. The children's original ROP status and neonatal cranial ultrasound scan data were extracted retrospectively from the clinical case records of the eight hospitals involved. As in our study, mothers not resident in the defined geographical region at the time of the infant's birth were excluded. A total of 143 infants weighed less than 1500 g at birth, and of these 115 had been examined for ROP. Thirty‐one were recorded as having had stage 1, 20 stage 2 and 8 stage 3 ROP. Infants over 1500 g birth weight were not routinely screened for ROP. At seven years, preterm infants were considerably more likely to wear glasses, to have poor visual acuity, reduced stereopsis and strabismus than term controls. The strongest associations in performance were between visual outcome and motor impairment. Ophthalmic impairments were considerably related to poorer scores on visual‐motor integration testing, the Movement ABC and Wechsler intelligence quotient (IQ) tests, with the exception of verbal IQ. Ophthalmic impairments were not significantly related to neonatal cranial ultrasound appearances. Stage 3 retinopathy was related to poorer subsequent acuity but not ROP stage 1 or 2. Our findings at secondary school (11–14 years) are therefore entirely consistent with this previous follow‐up in primary school at seven years. This adds weight to the general conclusions of both studies, particularly as our findings come from a different geographical cohort and time period with different investigators using different methods. The Liverpool cohort was selected by gestation and geographical area whereas ours was selected by birth weight and geographical area. Moreover, the ROP data used in our analysis were collected prospectively by a single examiner 21 and we did not routinely exclude children with major disability from ophthalmic follow‐up (although four were unable to perform the cognitive tests).
Our cohort had lower cognitive attainments at school age consistent with previous research.4,32,33,34,35,36 Our motor findings are comparable with another preterm cohort from a similar era.37 Roth et al,38 reporting outcomes at 14–15 years in a cohort born <33 weeks' gestation between 1979 and 1982, also found reading age was lower compared with standardised norms, but they had no detailed ophthalmic outcome data.
Our most striking finding was that cognitive attainment of low birthweight children at age 11–14 was more closely associated with their current visual status than with retinopathy in the neonatal period. Those with an adverse ophthalmic outcome scored, on average, 6.4 points less on spatial abilities (table 44).). It is biologically plausible that the spatial ability cluster of the BAS would be strongly associated with an adverse visual outcome. The association between BAS II scores and ophthalmic outcome remained after controlling for ROP, gestation, birth weight, cranial ultrasound, social class and sex. We do not have data on postnatal dexamethasone usage, which could affect both cognitive and visual outcomes by impairing grey matter growth.39 However, if postnatal steroids explained the association of BAS II scores and an adverse ophthalmic outcome, we would expect gestation and birth weight to be significant confounders, which they were not, as steroids were most frequently used in such infants.
Strabismus, abnormal three‐dimensional vision (stereopsis) and decreased corrected visual acuity were associated with BAS II spatial ability score (table 66).). Strabismus can also be due to cerebral palsy and neuromuscular impairments rather than ocular damage alone. However, irrespective of underlying mechanism, an adverse ophthalmic outcome results from lack of binocular functions plus amblyopia (reduced acuity).
It is sometimes not appreciated that a child can have reduced acuity without being myopic. The observed association of BAS II spatial ability with reduced acuity must be via visual pathway impairments rather than the refractory apparatus (cornea, lens, eyeball) because corrected visual acuity was assessed. Acuity was tested with spectacles (43/198 children wore glasses) so the reduced acuity was not due to myopia. Myopia was not associated with poor spatial ability. Finally, previous mild ROP does not explain these later problems with spatial ability (table 77).
Does the adverse ophthalmic outcome cause the poorer cognitive performance or are these associated because of damage to both the visual pathways and the cognitive centres? If the causal explanation is correct, it is possible that early identification of subtle visual disturbance may allow ophthalmic and educational intervention, concentrating on improving spatial skills. Against simple association is that visual problems at secondary school age are not explained by greater prematurity, lower birth weight or more abnormal neonatal ultrasound scan appearances (table 22),), all factors in poor cognitive outcome. 33,40 Similarly, when Cooke et al20 compared infants with birth weights above and below 1500 g, there were no significant differences in visual outcomes at seven year follow‐up. Furthermore, in that study, strabismus, poor visual acuity, absent stereopsis and low contrast sensitivity were also unrelated to the maximum extent of periventricular haemorrhage or periventricular leukomalacia on neonatal cranial ultrasound scan.20 However, the lack of a relationship between neonatal ultrasound scan appearances and the more subtle visual problems which we and Cooke et al20 describe do not exclude the possibility of a common hypoxic/ischaemic insult. Lower birth weight or major cranial ultrasound abnormalities37,40 can certainly predict major disabilities.32,34,41,42,43,44,45 Marked visual disability can also be predicted from cranial ultrasound abnormalities.46,47,48,49,50
However, in contrast to major disabilities, ultrasound prediction of learning difficulties, possibly due to less severe diffuse white matter damage,51,52,53 is less precise.54 The sensitivity of ultrasound to detect small areas of white matter abnormality is relatively poor.55,56,57 Furthermore, the effects of hypoxia/ischaemia may not be apparent on imaging until 32 weeks postconceptual age58,59 and may cause reduced grey matter volume60 which ultrasound cannot easily measure.
All but one of the subscales of the BAS II in which there were significant differences between the children with adverse and normal ophthalmic outcome involve a timed component. The education system places a great emphasis on timed procedures, such as school examinations. The potential additive effect of ophthalmic problems and underachieving in timed tasks might be substantial. Coupled with motor problems (which could cause poor handwriting), reduced reading comprehension and behaviour disorders,32 there are several additive factors which could lead to poor educational performance in ex‐preterm children without overt disability.
The main finding of the present study is that half of children born at less than 1701 g birth weight have an adverse ophthalmic outcome associated with poorer cognitive outcome at 11–14 years. Neither outcome is due to mild ROP. Rather, poor cognitive attainment is associated with adverse visual outcomes which manifest after the perinatal period. To determine causality, the effect of early detection and management of visual defects on later cognitive function could be studied. Low birthweight children should undergo early, regular visual assessment to detect visual impairment and enable intervention.
ABC - Movement Assessment Battery for Children
BAS - British Ability Scales
GCA - general conceptual ability
ROP - retinopathy of prematurity
This study was funded by Action Medical Research.
Competing interests: None.
Ethics approval was granted by University Hospital, Nottingham LREC.