Abstract
Objective
To establish whether the current vision screening practice in the Netherlands is effective in preventing permanent visual loss and to estimate the sensitivity of the programme.
Settings
In the Netherlands, all children are invited for preverbal (1, 3, 6–9 and 14–24 months) and preschool (36, 45, and 60–72 months) vision screening. Screening attendance is high, but the effectiveness in reducing amblyopia is unknown.
Methods
In a 7-year cohort study, 4624 children born in the city of Rotterdam between 16 September 1996 and 15 May 1997 were followed through all routine vision screening examinations. At age seven, visual acuity (VA) of children still living in Rotterdam was assessed by study orthoptists. In case of VA > 0.1 logMAR in one or both eyes, two or more logMAR lines of interocular difference or eye disorders like strabismus, children underwent a more intensive eye examination.
Results
Attendance at the 9-month screening was 89%, decreasing to about 75% at later examinations.
Of preverbal tests, 2.5% were positive, and of preschool tests, 10%. In total, 19% of children had a positive vision screening test at least once. Amblyopia prevalence was 3.4%. Sensitivity of the vision screening programme was 73% and specificity 83%. At age seven, 0.7–1.2% (confirmed vs final exam) of the children had a VA > 0.3 logMAR in the worse eye compared with 2–3.9% (in literature) reported prevalence in non-screening situations. Children who were less frequently screened had a higher chance of poor vision (>0.3 logMAR) at age seven.
Conclusion
The Dutch child vision screening programme may reduce the risk of persistent amblyopia (VA > 0.3 logMAR) at age seven by more than half.
INTRODUCTION
It is estimated that amblyopia occurs in about 2.7% to 4.4% of the general population. 1,2
Studies have shown that adults with amblyopia are at a greater risk for bilateral visual impairment. 3–7 If the function of the better eye is not affected later in life, the loss in quality of life is probably small 8 , but this affects a considerable portion of the population, as amblyopia treatment fails in approximately one third of cases. 9 The Australian Blue Mountains Eye Study showed that there was no significant difference in the distribution of the occupational classes between those with and those without amblyopia, but that significantly fewer people with amblyopia completed university degrees. 3 Rahi and colleagues found that major health and social outcomes were not significantly different between normals and amblyopes. 10 Such differences, however, are likely to be too small to be detected in a comparative analysis at the level of functional restrictions in daily life.
Treatment for amblyopia should, preferably, start before age six. The response to treatment at later ages is poor. 11 For this reason, it is important to detect amblyopia in early childhood.
This study followed a cohort of 4624 Rotterdam children – born between 16 September 1996 and 15 May 1997 – through all stages of the screening, referral, diagnostic and treatment process. In 2004, at age seven, all children underwent a final study examination by study orthoptists. The objective of the study was to establish whether the current vision screening practice in the Netherlands is effective in preventing permanent visual loss and to estimate the sensitivity of the programme.
METHODS
Since the introduction of the Child Health Care centres in the Netherlands, visual acuity has been measured in children from the age of three years. Visual acuity is assessed with the Amsterdamse Plaatjes Kaart (APK), the modified Amsterdam Picture Chart (APK-TOV), both non-validated, non logarithmic charts at age 36 months, or the standardized, logarithmic Landolt-C chart at 45–54 and 60–72 months (preschool screening). The APK is used instead if the child does not seem to understand Landolt-C testing. Vision screening is performed at child healthcare centres and public health services by specially-trained screening nurses and physicians. Preverbal screening of visual function before age three by means of the Vroegtijdige Onderkenning Visuele stoornissen (VOV) method (early detection of visual disorders for infants and toddlers) was introduced in the 1980's. 12,13 The VOV method consists of inspection of the outer eye, the pupillary light reflex test, assessment of strabismus by means of the corneal light reflex test and the cover-uncover test, and observation of the monocular (an indirect measure of visual acuity) and binocular pursuit movements performed by health professionals at 1–2, 3–4, 6–9, 14 and 24 months.
If an outcome is doubtful, because the child was, for example, uncooperative at that time, the test should be repeated within six weeks for preverbal screening and after three months for preschool screening. If the test outcome is abnormal (ie. the screening test is positive), the child is referred to an orthoptist or an ophthalmologist. This course of referral from screening physician to ophthalmologist or orthoptist usually occurs via the general practitioner (GP), who has the official gate-keeper role for most other health care services in the Netherlands. When a referral is made, the parents receive a letter referring them to their GP, requesting him/her to attend an ophthalmologist or orthoptist. This letter contains the results of the screening test. The GP may choose to re-examine the child's vision before making a further referral to an ophthalmology department. Nearly all orthoptists working in the Netherlands practice at ophthalmology departments in hospitals.
Included in the Rotterdam AMblyopia Screening Effectiveness Study (RAMSES) were all children born between 16 September 1996 and 15 May 1997 visiting the Child Health Care centres in Rotterdam from June 1997 onwards for their nine-month checkups.
14
In the Netherlands, 97% of all children of this age attend the child health screening programme, which is government paid. All Child Health Care centres and municipal health service locations in the Rotterdam region participated in the study. The original cohort consisted of 4624 children. By 2004, 949 children (21%) had moved outside the region and three children had died, leaving 3672 children for whom follow-up until the last screening examination was complete, and who were eligible for the final study examination. Of these, 81 children (2% of the total cohort) attended a school outside the region. For 423 children (9%) either the school was unknown or the school refused to participate in the final study examination; for 14 other children (0.3%) the parents gave no permission for the final examination. The remaining 3154 children (68%) were scheduled for the final study assessment (see flowchart). All parents were asked written permission by the health professionals.
We collected data about the eye examinations performed at the ages of 9, 14, 24, 36, 45–54 and 60 months from the Rotterdam Child Health Care centres and the municipal health service. To obtain information about the clinical follow-up, the orthoptists in the participating ophthalmology departments in Rotterdam and its suburban areas (Capelle aan den IJssel and Spijkenisse), were regularly sent lists of the names of children who had had a positive screening result and of the complete cohort (eg. including negative screening results). The orthoptists supplied data on orthoptic examination, diagnosis and treatment. The diagnoses made by the orthoptists were reviewed retrospectively by two independent experts (VKL and HJS), who were blinded to the screening results.
The diagnosis of amblyopia was upheld on expert review if it was likely that, with optimal correction, interocular acuity difference had been two or more logMAR lines at the start of treatment, no alternating fixation was present, and amblyogenic factors such as strabismus, anisometropia, astigmatism or deprivation could be identified.
At age seven, VA, stereopsis, eye motility, convergence and presence of strabismus were assessed in the final study examination. Study orthoptists visited 174 schools in the Rotterdam region to examine 2974 of the 3154 children (94%); 180 other children missed the examination (mostly due to holidays or illness). VA was measured at a distance of 5 meters using mobile light panels with a transparent, logarithmic Landolt-C chart. The child was invited for a more extensive orthoptic examination at the paediatric ophthalmology department at Erasmus MC-Sophia in Rotterdam if one of the following abnormalities was found: VA > 0.1 logMAR in one or both eyes; two or more logMAR lines of interocular difference; manifest strabismus; heterophoria; having failed TNO random-dot stereopsis testing or stereopsis ≥ 240 seconds of arc; or impaired eye motility.
This extensive orthoptic examination consisted of automated refraction and retinoscopy to determine refractive errors, including astigmatism. Retinoscopy was repeated after cycloplegia (cyclopentolate 2%) if any significant refractive error was suspected. VA was measured with best refractive correction. As a separate additional independent test, stereopsis was determined with Titmus, LANG and TNO-RDT tests. Manifest or latent deviations were measured with the prism cover-uncover test and the alternating prism cover test, at near with a fixation object and at distance with a fixation light. Finally, the child's eyes were examined by slitlamp and funduscopy by an ophthalmologist (HJS).
Endpoint was the best-corrected VA in the worse eye at age seven. Visual impairment was defined as a VA > 0.3 logMAR in the worse eye (poor vision) or a VA > 0.5 logMAR in the worse eye (legally blind).
Sensitivity and specificity
Sensitivity was defined as the proportion of children with amblyopia who had a positive vision screening result at any point in time. It represents the ability of the Dutch child vision screening programme to detect cases of amblyopia. Specificity was defined as the proportion of children without amblyopia who had no positive screening result. At age seven before the final study examination, all parents were asked if their child had ever visited an eye specialist/orthoptist (independent of the screening result).
Statistical analysis
Data were analysed in SPSS version 15. Differences between subgroups of children according to visual acuity were tested with Fisher's exact test. The level of significance was set at p = 0.05.
RESULTS
The initial screening response rate at 9 months was 89%; this decreased to about 75% at later screening examinations (figure 1). Of the 3672 children for whom screening FU data were complete, 3569 (97%) had undergone preverbal screening using the VOV method at least once; 3522 children (96%) had undergone screening with visual acuity testing at least once. For 26 children (0.7%), no screening results were available. For approximately two-thirds of children (2506 of 3672), five or six out of a maximum of six screening results were received.
Number of screening examinations and results for 3672 children with complete follow-up until the last screening examination at age 60 months
One or more positive screening results were found in 711 children (19%) (figure 2). Of these children, 155 children (22%) had their first positive screening result at preverbal screening, the other children at preschool screening using visual acuity testing. For 407 of the 711 children with a positive screening result (57%), data were available from the orthoptist or ophthalmologist who had treated them. Another 90 children with a positive vision screening test had visited an orthoptist or ophthalmologist, according to their parents.
Number of children with a (first) positive screening result by screening examination
At age seven, 2964 children underwent the final study examination. Ten other children were excluded from this analysis, because it had not been possible to assess their VA with Landolt-C optotypes.
Because of the findings at the initial orthoptic examination at school, 407 children had been invited for an extensive ophthalmologic and orthoptic examination; 83 of these (20%) did not participate: 23 parents refused because their children were already under orthoptic care and seven children had moved. The reason for non-participation was unknown in the other cases.
Children who underwent the final assessment and those of 3762 who did not, did not differ significantly concerning gender (boys 49% versus 51%; p = 0.4), being screened positive (19% versus 21%; p = 0.2) or known treatment by orthoptist or ophthalmologist (13% versus 12%; p = 0.3).
Visual acuity in the worse eye at age 7 (final study examination)
Screening frequency and chance of visual acuity > 0.1 logMAR and > 0.3 logMAR at age 7 *
* Data on visual acuity in the worse eye at age 7 were missing for 2 of 2964 children (see table 1); these children were excluded from the analysis. P-values concern the comparison between screening characteristics and the proportion of children with (best corrected) visual acuity > 0.1 logMAR (> 0.3 logMAR, respectively) at the initial or additional orthoptic examination, using Fisher's Exact test
† Data on 13 children of the 2962 children are incomplete
Screening performance
Clinical characteristics and visual acuity at age seven of children with confirmed amblyopia (n = 100)
* Excluding the 17 children without any orthoptic or ophthalmologic care before age 7.
† Five of 9 children receiving occlusion therapy underwent surgery as well, and 9 of 55 children receiving both occlusion therapy and glasses
DISCUSSION
In this birth cohort, most children were intensively screened. The overall prevalence of amblyopia in our cohort was 3.4%. At age seven, 23 of the 100 amblyopic children had visual acuity > 0.3 logMAR. Overall, 1.2% of children had VA > 0.3 logMAR at age seven. This prevalence of poor vision is in accordance with prevalences found in screened and treated populations. As a randomized study design was not possible, there was no control group in the RAMSES study. The number of screening examinations per child, however, was inversely correlated with visual acuity at age seven: children who had been screened less intensively were more likely to have poor vision.
We also saw an important contrast with children who were not screened until after their third birthday. However, no adjustments could be made for the less favourable prognosis of undetected amblyopia during vision tests beyond the age of four. Although the loss of children in FU may have introduced a bias, it seems limited as the majority was related to moving outside of the region. At the same time, our measures of screening performance will be over-optimistic because all data are based on children attending some kind of health assessment. There will be children with undetected amblyopia in the community that never present to the health service system.
A cohort study among 6081 British children showed that the prevalence of persistent amblyopia at age 7½ was significantly lower among children who had received screening at age 4–5 than among children who had never been screened (1.1% vs. 2.0%). In that study, amblyopia was defined as two or more logMAR lines of ocular difference. 15 A visual acuity of < 6/12 in the worse eye was present in respectively 0.7% and 1.3%. In a randomized controlled trial among 3490 children, the same researchers compared intensive orthoptic screening at the ages of 8, 12, 18, 25, 31 and 37 months with screening only once at the age of 37 months. 16 In that study, 0.6% of the children in the intervention group had VA < 6/12 in the worse eye versus 1.8% of the children in the control group. Moreover, the visual acuity in the worse eye of amblyopic children in the first group was better than that of amblyopic children in the second group.
A comparison of the observed prevalence of persistent amblyopia in our study (ie. 0.8%) with the prevalence in literature, 2–3.9% 17 , suggests that more than half of potential vision loss due to amblyopia at age seven may have been prevented by the programme in the Netherlands.
The cost of visual screening within the scope of the primary health care service is an estimated € 4 million. Whether screening for amblyopia is cost effective is undetermined, as little is known about the effect of amblyopia on quality of life in the short and/or the long term. 18,19 The group of amblyopic children is a heterogenic one. Not all undiagnosed amblyopes will suffer substantially from their disorder later in life, whereas not all (early) detected children may have needed the treatment. A longer follow up of the children and a ‘best guess’ of who may and may not have benefited, but should have benefitted, is recommended. 19 We realize that the frequency of vision screening in the Dutch child health monitoring programme is far in excess of what is carried out in many countries (eg. the UK). In the 1980's, preverbal vision screening was added to the Dutch child health monitoring programme, as physicians thought that screening of vision in infants and toddlers, like hearing screening (using the Ewing test), would be useful. For that purpose, the VOV-method was developed as a vision screening method in very young children. However, an intensive vision screening programme consisting of 6–8 examinations, as is currently integrated in the Dutch child health monitoring programme, may not be cost-effective, and the marginal cost of adding this vision screening programme to the already existing child health monitoring programme has been limited. The cost-effectiveness, therefore, is likely to differ substantially when implementing this programme from scratch and on a large scale. There is considerable debate over whether fewer screens, one screen only beyond neonatal screens, or no screen at all would be required. In our study, we found a negative association between the number of screens received and the presence of persistent amblyopia at age seven. The relatively high number of false-positive screens, however, increases cost substantially. For that reason, we are starting research on the (cost) effectiveness of a less intensive vision screening programme. Using a cost-effectiveness model, the benefits of early detection and treatment can be calculated for each type of amblyopia. The results of this analysis will reveal if the child vision programme should be altered in future. It has to be noted that the relatively low sensitivity at younger age (VOV-method) is indeed under the assumption that all diagnosed cases of amblyopia were already present at very young age. The gold standard is more difficult to establish at young age.
Furthermore, it is likely that the screening technique is more difficult to apply at the young age when the child may be more distracted, compared with a screening technique at school age, and the criteria for referral are more subjective than at older age.
In our study, children with strabismus or combined amblyopia were about 16 months younger than children with anisometropic amblyopia when they first visited the orthoptist or ophthalmologist. 20 Thirty-five years ago, the various types of amblyopia were detected around two years later, though with the same age sequence. 10 In that historic cohort study in the late 1960's, occlusion therapy was started at a mean age of 5.1 years for strabismic amblyopia, 5.7 years for combined-mechanism amblyopia and 6.6 years for anisometropic amblyopia. In a retrospective Australian study of 127 children with amblyopia, a trend was also seen for earlier detection of deprivation amblyopia and later detection of anisometropic amblyopia, though amblyopia type and age at the first outpatient visit were not significantly related. 21
As wide scale screening for amblyopia had already been introduced in the Netherlands, no randomized controlled trial study design was possible. Nevertheless, our results seem to be in agreement with the UK-study. We assume that the low(er) percentage of persistent amblyopia at age seven is largely owing to the vision screening programme implemented in the primary health care service.
Footnotes
ACKNOWLEDGEMENTS
The following people and institutions contributed importantly to this study:
– the physicians and health care professionals at the well-baby clinics of Stichting Thuiszorg Rotterdam and at the greater Rotterdam municipal health service; – the orthoptists and ophthalmologists at the Rotterdam Eye Hospital, Erasmus MC, Erasmus MC Sophia, Ikazia Ziekenhuis, Medisch Centrum Rijnmond Zuid – locatie Zuider, Medisch Centrum Rijnmond Zuid – locatie Clara, St. Franciscus Gasthuis, Ruwaard van Putten Ziekenhuis and the IJsselland Hospital; – data managers Michel Provoost, Roel Faber and Jan Alberts at the Department of Public Health, and René Boer of the greater Rotterdam municipal health service; – research assistants Ilse Oonk, Annemieke van Eijsden, Priscilla Dorresteijn and Mirjam Goudswaard at the Department of Public Health; – the schools in Rotterdam; – the study orthoptists Annemarie Uijttewaal, Carolien ten Wolde-van Voornveld, Aysegül Tonkul, Babes Klug-van Veen, Liesbeth Ploeg, Inez Elderman, Esma Aygun, Ellen van Minderhout, Brigitte Simonsz, Jan-Roelof Polling, Marleen Vermeulen, Isa Vels and Jolanda de Vos-van den Broek; – Paul J. van der Maas, MD, PhD; – Arry E. de Bruijn.
This study was financed by Praeventiefonds / ZonMW [registration number 6320.008].