Article Text

Download PDFPDF

Are there inequities in the utilisation of childhood eye-care services in relation to socio-economic status? Evidence from the ALSPAC cohort
  1. M Majeed1,
  2. C Williams2,
  3. K Northstone3,
  4. Y Ben-Shlomo1
  1. 1
    Department of Social Medicine, Canynge Hall, Bristol, UK
  2. 2
    Academic Centre of Child and Adolescent Health, Bristol, UK
  3. 3
    Department of Social Medicine, ALSPAC, Bristol, UK
  1. Ms C Williams, Department of Community Based Medicine, 24 Tyndall Avenue, Bristol BS8 1TQ, UK; cathy.williams{at}


Background: Equity of access to eye care in childhood remains poorly researched, and most studies report data on utilisation without any objective measure of clinical need.

Participants/method: 8271 participants from the Avon Longitudinal Study of Parents and Children (ALSPAC), a longitudinal birth cohort, were seen at age 7, when they underwent a comprehensive eye examination and details of family history of eye conditions, vision problems and contact with eye-care services were obtained.

Results: 2931 (35.4%) children had been in contact with an eye-care specialist, and 1452 (17.6%) had received vision screening. Compared with social class I, the prevalence of eye conditions was higher in the lower groups (social class IIIM, IV, V) (OR 1.69, 95% CI 1.15 to 2.46). However, children from lower socio-economic status groups were less likely to see an eye-care specialist (OR 0.83, 95% CI 0.70 to 1.00) or to use screening services (OR 0.65, 95% CI 0.43 to 0.98).

Discussion/conclusion: The differences in the trends between socio-economic groups in eye conditions and utilisation of services suggest inequitable access to services. These data highlight the limitations of community-based preschool vision screening, which fails to abolish this inequity. It is important that future research explores the reasons behind these patterns. Compulsory school-entry vision screening, as recommended by the National Screening Committee and the Hall Report may redress this differential uptake of services.

View Full Text

Statistics from

The importance of a screening and surveillance programme for childhood visual problems was laid out by the World Health Organization initiative “Vision 2020.”1 2 The The UK has successfully implemented various programmes over the past few decades, and visual screening has been widely but variably available in the UK since the 1970s.3 There is currently an intense debate about the use of population versus targeted vision screening and its effectiveness in reducing visual problems in children at population levels.47 Parents also obtain access to eye care through a variety of sources, including high street optometrists, community-based orthoptists in health centres and specialist eye-care centres based at hospitals. Relatively little is known about who uses these services in relation to socio-economic factors. Thus, it is not clear whether the children who access eye services are actually those who would benefit from early detection and treatment. Any mismatch between provision of clinical services and clinical need is regarded as evidence of inequitable access to healthcare.8

The prevalence of paediatric eye conditions is common and estimated to be 2.4–12.8%,6 912 based on various definitions. Very little information is available on how morbidity varies across socio-economic groups. Data from the Avon Longitudinal Study of Parents and Children (ALSPAC) have highlighted an increase in the prevalence of some childhood eye problems (hypermetropia convergent squint and amblyopia) in lower parental social classes so that children from poorer families have a greater burden of morbidity.12

This study aims to analyse the variations in the utilisation of childhood eye-care services in relation to socio-economic status and examine whether utilisation patterns are equitable in relation to clinical need as assessed using objective measures. We hypothesised, a priori, that if children from deprived families had a greater burden of eye disease, then they should have concordant higher rates of service utilisation.


Study participants and data collection

ALSPAC is an ongoing birth cohort study of children born to mothers with an estimated date of delivery between 1 April 1991 and 31 December 199213 in the area that was the county of Avon, in Southwest England, UK. Children attended a research clinic at around 7 years of age, where they underwent a 20-minute visual examination with an orthoptist testing monocular visual acuity, ocular alignment and autorefraction. This clinic began in October 1998 and ended in October 2000.

Measures of service utilisation

Parents provided information on previous utilisation of eye services via a self-completion questionnaire, which was then checked at the clinic. Information was obtained about the professional, for example optometrist or ophthalmologist, and the setting, for example eye hospital or health centre. These data were classified into four groups; any eye specialist, optometrist, health centre or hospital/ophthalmologist (including the private sector). Further data were obtained by record linkage with existing health-service records to determine whether children had participated in a routine preschool orthoptic vision screening service (“state screening”). This service was only offered to a subset of the participants who lived within one of the three health districts which comprise the study area. All were offered standard school-entry vision screening by school nurses.

Measures of socio-economic status

Demographic data were collected during the 32nd week of pregnancy. Mothers’ and partners’ social classes were classified according to The 1995 Standard Occupational Classification System14 based on current occupation. This is a six-level ordinal variable where a lower value indicates better socio-economic status. The more advantaged of the two parents’ social classes was used to define “parental socio-economic status” (SES). The numbers of subjects from social classes IIIM, IV and V were relatively small so a four-level ordinal variable was created (I, II, IIINM, IIIM-V). Maternal level of education (<16 years (CSE/vocational), 16 years (O level), 18+ (A level or higher)) and housing tenure (owned/mortgaged, private rented, council/Housing Association) were also considered as complementary measures.

Clinical measures of vision problems in childhood

Vision testing at age 7 years was carried out by orthoptists working to a specific protocol and included, but was not limited to, the following assessments. The presence of strabismus was tested for, using the cover/uncover test and the alternate cover test, both conducted at near (33 cm) and distance (6 m) fixation, with and without glasses, if worn. Any deviation was measured with the simultaneous prism cover test. Deviations were classified according to direction (convergent, divergent, vertical, mixed) or duration (intermittent, constant). We classified any manifest (present on cover testing) deviation of 2 pd or greater, at either distance with or without spectacles as “manifest strabismus.”

Refractive status was assessed using a Canon R50 (Canon, Tokyo) autorefractor, without cycloplegia. A nested validation study was conducted to determine the reliability of these data in comparison with the results of cycloplegic retinoscopy15. The autorefractor was 71% sensitive and 99% specific when used to identify children with optometrically confirmed hypermetropia. Hypermetropia in this analysis was defined as a spherical equivalent of ⩾+2.00 D in either eye (severe hypermetropia if spherical equivalent was ⩾+4.00 D).

Visual acuity was tested monocularly using ETDRS charts, in their habitual state (ie, with glasses if worn) and then habitual state plus pinhole. The better of these two measures for each eye was taken as their best-corrected acuity. Amblyopia was defined as previous history (parental report) of any patching of the better-seeing eye, or difference in best-corrected visual acuity (tested as above) between eyes of >0.2 logMAR units, with the worse eye having a visual acuity of >0.3 log MAR units and no ocular pathology seen on dilated funduscopy. A composite outcome variable was derived, called “any vision problem,” which combined hypermetropia, convergent squint or amblyopia. Children with significant non-refractive ocular co-morbidity were excluded from the study.

Other confounding or intermediary variables

We hypothesised that family history of eye problems and lay perception of eye problem may also determine healthcare utilisation and may be related to SES either as confounders or intermediary variables. The questionnaire included questions on a family history of “squint” and/or “lazy eye.” This was classified as either (1) any family member or (2) only other siblings had an eye problem. The latter subgroup may be more pertinent as the parent has had a previous experience of eye problems and hence may have a lower threshold for seeking eye care. Parents were also asked if they, their partner, a relative or friend had ever thought that the study child had an eye problem (“lay perception”).

Statistical methods

Chi-squared tests were used to assess the association between SES and patterns of service utilisation. Odds ratios (ORs) and 95% confidence intervals (95% CI) were calculated using logistic regression. As SES is an ordinal variable, we examined the association both as a dummy variable and a linear term so that we also present the odds ratio for a unit change in SES and p value for trend. We undertook a series of multivariable logistic regression models; (1) adjusting for age and sex, (2) as above but also adjusting for family history of eye problems and lay perception of eye problem and (3) as above but also adjusting for prevalence of eye disease. Analyses of uptake of state screening were based only in the subset of children who were invited to attend. Because some of these children were also screened as part of a nested randomised controlled trial within ALSPAC,16 we repeated these analyses excluding these children as their parents may have chosen not to attend preschool vision screening because they had already been screened.


From the 14 541 pregnancies recruited, 13 988 children were alive at 12 months of age, and 8290 attended the visual assessment clinic at age seven (56.0% of those alive at 12 months, plus another 455 who joined the study after infancy). Nineteen were excluded due to serious non-refractive ocular co-pathology, leaving 8271. Data on service utilisation were missing for 558 (6.8%) children, while the SES of the parents was recorded for 7186 (86.9%). The baseline characteristics of the sample are shown in table 1. There were roughly equal numbers of boys and girls. The majority of parents were from social class II, with a third having continued education beyond 18 years (A levels), and 75% lived in their own homes. There were very few non-white subjects (1.8%). Around a third (35.4%) of children had some contact with an eye-care specialist, but only 9.4% (n = 777) had been seen at an eye hospital or by an ophthalmologist. A quarter (n = 2025) of children had been invited to a preschool vision screening programme; 72% (n = 1457) actually attended, though 110 children were already attending clinics elsewhere; hence the proportion of children not already known to services taking up screening was 66.5%. Eye conditions were found in 7.5% (n = 539); hypermetropia was the most common condition. In contrast parents reported much higher rates of both family history (23.6%) and a non-medical person (either parents, family, friends or school teacher) noticing a vision problem in the child (12.1%).

Table 1 Baseline characteristics of the study subset of ALSPAC Cohort, with data on sociodemographic, clinical and other variables that could determine utilisation of eye services

There were variations in healthcare utilisation and morbidity of eye diseases in relation to social class (table 2). Children from deprived families were generally less likely to have sought care, although the relative differences were modest. There was a trend between SES and visits to eye-care specialists and preschool vision screening such that lower SES was associated with decreased service use. The association with preschool vision screening was more marked when the subsample involved in the nested RCT was excluded. Similar findings were seen when we used maternal level of education and housing tenure as our measures of SES. For example, children whose parents lived in council or housing association accommodation had an odds ratio of 0.62 (95% CI 0.45 to 0.87) for attending screening compared with owner-occupiers. In contrast, the prevalence of eye disease showed a positive association between objective vision problems and SES such that children from deprived families had a greater burden of need. This was most marked for hypermetropia (OR 1.69, 95% CI 1.09 to 2.61).

Table 2 Adjusted* odds ratios (95% CI) showing variations in the utilisation of eye services and prevalence of eye disease with parental social class

Both a positive family history (odds ratio 1.84, 95% CI 1.66 to 2.05, p<0.001) and a lay perception of a vision or eye problem (odds ratio 25.1, 95% CI 19.8 to 31.8, p<0.001) predicted consultation with an eye specialist but were not related to uptake of state screening. The diagnostic utility of a lay perception of eye problem was modest (positive likelihood ratio 4.2) as though the specificity was good (89.4%), and the sensitivity was less impressive (44.6%).

We also examined the association between other socio-economic measures such as maternal level of education and housing tenure. These showed, if anything, weaker associations except for housing tenure with uptake of screening (odds ratio council versus own occupier 0.62, 95% CI 0.45 to 0.87, p = 0.005).

Table 3 shows three models of analyses for service utilisation. In order to directly compare the odds ratios, we only kept the subset of subjects with complete data for all the co-variates (n = 5216), despite some loss of statistical power, so that differences between models reflect the effects of adjustment and not loss of observations. This is why the results in table 2 are not exactly the same as the results for Model 1 in table 3.

Table 3 Multivariable associations between utilisation of eye services and social class*

For all the outcome measures, the association with SES was inverse and in general was strengthened after further adjustment. The strongest associations were seen with consultation with any eye-care specialist (model 3 p = 0.007) and uptake of state screening, excluding those subjects that were in the nested RCT (model 3 p = 0.06). The association with SES was generally linear, in a dose–response fashion, except for attendance at a health centre, where the odds ratio for social class IIIM was greater than for social class II. The effects of social class II for state screening were similar to that seen with the baseline group (social class I).


This study shows that the prevalence of the conditions we studied is higher among children in the lower social classes, suggesting they have a greater healthcare need. Paradoxically, they access services less than those from higher socio-economic groups. This is true for both services that require individuals to initiate the consultation, for example optician, and for screening services where individuals are invited to attend. This mismatch between service utilisation and clinical need highlights inequities in access of services. This analysis is cross-sectional, but it may be possible that the lower prevalence of children with vision defects at 7 in the higher SES groups is due to previous greater service usage by these families, resulting in better treatment and/or prevention of vision problems. Further, longitudinal data are needed to explore this possibility.

Of the study population, 7.5% reported or had some form of vision condition. The prevalence of refractive problems in our sample was lower than those reported in most parts of the world.6 810 1720 These variations appear to be partly due to the fact that we considered only one refractive error (hypermetropia), and the wide differences in the definitions of different refractive disorders. From the study population, 37.4% had reported seeing an eye-care specialist. This figure is similar but higher than a similar study in Australia.21 No other published data are available for comparison. As well as SES, we also noted that a positive family history and lay concerns about eye problems predicted consultation. These results are consistent with a previous case-control study which reported that the same factors were associated with early as compared with late diagnosis of amblyopia.22 The reduced uptake of screening by children from lower SES groups has also been reported from a smaller study of children from Glasgow.23 Overall our results of social inequities in uptake of eye-care services are consistent with studies in other disease areas or different screening programmes.2428

Limitations and scope for further research

ALSPAC is a single birth cohort, and the participants come from a defined geographical area. The proportion of non-white families was disproportionately small (1.8%) compared with the general UK population (7.9%).29 Therefore, our results may not be generalisable to other geographical areas or ethnic minority groups, though we do not believe that availability of eye services differ markedly across the UK.

We limited our definition of visual conditions to only include hypermetropia, amblyopia and convergent squint, and did not include astigmatism and myopia (which are less well identified by the study methods we used). The refractive data were collected without cycloplegia, and so we are likely to have underestimated the true prevalence of hypermetropia and so reduced our ability to quantify associations between hypermetropia and SES or service usage. Our data do not allow us to understand the reasons for low uptake of services among children from poorer backgrounds. Future qualitative research may help understand to what degree sociocultural attitudes or barriers exist that affect the behaviour of parents in using eye-care services.

Public health implications of the findings

Reports from this cohort and others have shown benefits of intervention in treating childhood eye problems.3032 Our results demonstrate a significant difference between the socio-economic groups in their pattern of service use. It is difficult to establish whether these differences reflect underuse of services by children from lower SES or overuse of services by children from, higher SES.

The concept of universal preschool vision screening has been the subject of intense debate in recent years. It is no longer recommended in the UK for the purpose of early detection of eye disease, as the reduction in the prevalence of amblyopia at age 7 is only reduced by a modest degree.30 The Hall report33 acknowledges this gap in the care received by children from lower socio-economic backgrounds. It not only calls for set referral pathways for vision problems in childhood, but also recommends universal vision screening at school entry as opposed to preschool, as this results in better coverage. Even if this is achieved, it will be important to ensure that all children have access to optimum care subsequently. Whether such a policy will fully redress current inequities remains a question that future research needs to answer.


We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes orthoptists, interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. The UK Medical Research Council, the Wellcome Trust and the University of Bristol provide core support for ALSPAC.


View Abstract


  • Funding: The Research and Development Directorate of the Southwest Regional Health Authority specifically funded this collection of the vision data, presented here.

  • Competing interests: None.

  • Ethics approval: Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and the Local Research Ethics Committees.

  • Patient consent: Patient consent was obtained.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Harminder S Dua Arun D Singh