Results

Of 413 VLBW infants admitted to New Zealand neonatal units in 1986, 338 (81.8%) survived to discharge home and of these 313 (93%) had been examined for acute ROP. Twelve children died after discharge from hospital and before 7 years of age; 17 children (5.2% survivors) were traced to overseas; the families of seven children (2.1% of survivors) declined to participate in the study; and four children (1.2%) of survivors remained untraced. Hence, 298 children (96.4% of surviving children resident in New Zealand; 91.4% of all survivors) had some assessment but two of these children did not have visual assessment having moved overseas before this was carried out. The mean age at assessment was 7.6 (SD 0.44) years.

ROP was present in 66/313 infants (21%) and absent in 247. Of 66 children with acute ROP (ROP+), 64 survived to age 7–8 years and 58 (91%) were assessed. Among these 58 children, the highest stage of acute ROP was stage 1 in 26, stage 2 in 20, stage 3 in seven, and stage 4 in five. Of 247 children examined for but without acute ROP (ROP−), 238 survived to age 7–8 years and 216 (91%) were assessed; and of 25 children not examined for acute ROP (NA), 24 survived to age 7–8 years and 22 (92%) were assessed.

Five children were blind from ROP and comprehensive information on their vision was available from the New Zealand School for the Blind. Six children (one with cortical blindness, one with severe cerebral palsy, one with autism, and three with attention deficit disorder) could not be tested on distance vision. Six children (one with autism, five because of camera failure) were not tested by photorefraction. In only two children was there no information either from distance vision testing, photorefraction, or other sources.

Eight children were registered blind (vision less than 4/40: Table 1). All five children with grade 5 ROP had a birth weight less than 1000 g and gestation less than 28 weeks (mean 778 g; 25.8 weeks). The incidence of blindness overall was 19.3 per 1000 live births with birth weight less than 1500 g. Of the seven children whose parents declined to participate in the study, one (also with birth weight less than 1000 g and gestation less than 28 weeks) was known to have bilateral blindness from ROP, and if included would increase the incidence of blindness to 21.7 per 1000 live births.

Overall, 64% of children had some visual abnormality with significantly more children who had had acute ROP having a visual problem than children who did not have acute ROP (Table 2). (In all cases children who were not examined for ROP had rates of abnormalities that were similar to, or lower than, those of the ROP− children and so have been included with this group). There were no differences in rates of hypermetropia between the groups and if this is excluded, the overall rate of visual abnormality reduces to 56%.

By photoscreen assessment, mild (0.5–2.00 D) and moderate/severe (> 2 D) myopia in the better eye occurred in 8% and 17% of ROP+, and 10% and 1% of ROP−/NA children (p < 0.001). In the worse eye, figures were 19% and 17% of ROP+, and 15% and 3% of ROP−/NA (p < 0.001) respectively.

We were unable to clearly identify which children had amblyopia because we did not examine the fundi to exclude retinal disease as a cause of diminished visual acuity. However, amblyopia may be present where there is strabismus (which was significantly more frequent in ROP+ children; Table 2); where distance visual acuity varied by more than two levels between eyes (ROP+ 10.3% versus ROP−/NA 7.6%: NS); or where this latter category is combined with strabismus (ROP+ 3.5% versus ROP−/NA 4.6%: NS).

The highest stage of acute ROP was significantly associated with subsequent visual outcome. Any visual abnormality was seen in 60% children in both ROP− and NA groups, 62% of those with stage 1 ROP, but 90% with stage 2 and 100% of those with stage 3 or 4. The visual acuity for individual eyes related to the highest stage of acute ROP is shown in Table 3.

Discussion

In this prospective population based study of VLBW infants we have detected a high incidence of ophthalmic abnormalities at 7–8 years of age; nearly two thirds of the cohort having some abnormality and one in five children having spectacles prescribed. The presence of acute ROP of any stage in the neonatal period was associated with a near doubling of poor distance and near visual acuity, photoscreen myopia, and strabismus compared with rates observed in infants with no acute ROP. Rates of significant hypermetropia, astigmatism, and poor stereopsis were no different between the groups.

We chose to visit children in their homes in order to achieve the highest possible follow up. We also used non-medically qualified home visitors who were skilled in using the WISC-R. Although these visitors were trained in the techniques of visual assessment, this meant that fundal examination was not carried out and hence we were unable to distinguish between cicatricial and regressed ROP. Additionally, photorefraction was used which should detect all refractive errors of more than 2.00 D but may miss some more minor degrees of refractive error. If anything this will mean we may have underestimated the rate of visual abnormalities, although we are unlikely to have failed to detect significant functional problems.

The present study also did not include a contemporary control group of infants born at term because to do so would have greatly increased the cost and presented numerous logistical problems. Comprehensive information on vision and eye problems at age 7 in New Zealand children is available from a study of 988 Dunedin children born in 1972/73, which cohort included no VLBW children.16 In this group any visual abnormality occurred in 9.4%. Distance visual acuity of less than 6/12 occurred in the better eye in 2.2% and worse eye in 5.5% of children. Near visual acuity worse than N5 occurred in the better eye in 1.0% and worse eye in 4.4%. Strabismus occurred in 3.9% and 3.6% of children had had spectacles prescribed. There is nothing to suggest any change in the overall incidence of childhood visual disorders in New Zealand in recent years. Comparing our data with these Dunedin data it is clear that VLBW, even in the absence of ROP, does greatly increase the risk of visual disorders by 7 years, and for important outcomes by two to three times.

There have been few population based studies which have comprehensively assessed visual outcome in VLBW infants related to the presence or absence of acute ROP,17-20 and the differing study design and method of reporting makes comparison with this study difficult. The study from the East Midland Region of the UK is perhaps closest to ours in design21 but has not reported on visual outcomes beyond the first year of life.19 The important population based cohorts from Canada22 and Australia23included only infants with birth weight less than 1000 g and focused primarily on overall neurodevelopmental status. Much information is becoming available from the natural history arm of the large multicentre CRYOROP study72425 but this study only included infants with birth weight less than 1251 g and needs to be interpreted alongside population based data.

Although our rate of visual abnormalities is high, other studies have also found high rates of visual problems in ex-very premature children. A report from the Oxford Region found 59% of surviving infants with birth weight less than 1000 g had ocular abnormalities at 3–5 years of age.18 A study from Sweden reported 25% of children born < 33 weeks’ gestation or with birth weight <1501 g in Stockholm County had been referred for ophthalmic care aged 5–10 years, compared with 11.5% of term children.2627 A number of unit based studies report similar data. Dowdeswell and others found some ocular pathology at age 5–7½ years in 45% of surviving infants with gestation less than 32 weeks, compared with 13% term controls from the south west of England.28 Two studies from the Netherlands have reported visual impairments in 54% of VLBW infants at 1 year of age and 45% of premature infants at 6–10 years of age.329 With the exception of the study by Cats and Tan,3 appropriate screening for acute ROP was not carried out in these studies, hence no inferences can be made about the contribution of ROP to visual outcome.

It is well recognised that refractive errors, amblyopia, and astigmatism are common in premature and very low birthweight infants whether or not they have suffered acute ROP.5630Approximately 90% of cases of acute ROP will spontaneously regress31 but it has been unclear whether completely resolved ROP is associated with any visual sequelae, or alternatively whether visual problems in premature infants may be the result of unrecognised ROP.

Many studies have reported increased rates of myopia in children who have suffered acute ROP.3573032-37 Recent studies with follow up beyond 1 year of age and in which infants had been screened for acute ROP are shown in Table 4. These studies all show approximately twice the rate of myopia following acute ROP compared with absence of ROP; the overall rates in both groups being similar to the present study.

It has been suggested that myopia of prematurity tends to regress during the first year of life resulting in emmetropic or hyperopic refractions later, but when ROP develops this shift does not occur.3 If this is what takes place there may be an expectation that rates of hypermetropia would be lower in infants with ROP compared with those without. However, both in the present study and that of Cats and Tan (who reported 11% hypermetropia ⩾ 1 D overall)3 there was no such difference between the groups. It may be that milder degrees of ROP, in terms of both stage and location, have less effect on ocular growth, as has been suggested by both the CRYOROP study7 and the East Midlands study.19 The relation between prematurity, acute ROP, ocular growth and development, and subsequent refractive index has been discussed4719 and possible contributing mechanisms include influences on corneal curvature,38 lens thickness,39 anterior chamber depth,40 as well as axial length.

While it is clear that more severe ROP is associated with overall worse outcome it remains controversial as to whether regressed stage 1 and 2 disease is associated with visual sequelae. Data from the CRYOROP study confirm that serious adverse outcome (defined as a retinal fold involving the macular, retrolental opacity encroaching on the pupil or partial or complete retinal detachment) is rare after stage 1 or 2 disease other than in zone 1.25 On the other hand in that study, stage 2, zone 2 disease was associated with nearly twice the rate of subsequent myopia compared with no ROP, as well as an increased risk of strabismus.725 Our data would support this view (Table 3), with stage 2 disease, but not stage 1, increasing the risk of mildly reduced visual acuity. The implication of this is that any child who has had acute ROP should be followed at 1 and 2 years of age to check acuity and refraction as a minimum.

We found an overall rate of strabismus of 22%, similar to that observed in other studies,318353741 although Page and others36 reported lesser rates at 12 and 24 months of age. Most studies do suggest strabismus is nearly twice as frequent after acute ROP compared with similar infants without ROP.32535 Only Robinson and O’Keefe found a higher rate in infants without ROP and there is no clear cut explanation for this.37

There has been some debate as to whether the use of antenatal steroids, given to reduce the risk of hyaline membrane disease, may be associated with a reduction in the incidence of blindness from ROP such that cortical blindness is now the more common cause of blindness in premature infants.42 Since the pioneering work of Liggins and Howie43 New Zealand has had high rates of antenatal steroid usage and 60% of this cohort received a full or partial course in utero. Cryotherapy was not available in New Zealand in 1986 but had it been, the number of eyes with adverse visual outcome may have been reduced by nearly 50%, which may have meant children with blindness from ROP were either the same number or fewer than those with blindness from other causes (principally cortical blindness). Given that there are likely to be some children with cortical blindness, but not blindness from ROP, who had birth weight greater than 1500 g, it does seem plausible that blindness due to the former may exceed that due to ROP. However, this study suggests that if there is a decline in blindness from ROP it is more likely to be due to appropriate screening and treatment with cryotherapy than to antenatal steroid usage.

In conclusion, despite a number of unavoidable limitations in our study design, our data provide further evidence that VLBW greatly increases the risk of childhood visual disorders, and that a history of ROP is associated with some additional risk, including a near doubling of reduced visual acuity, myopia, and strabismus.