Objectives: To evaluate different refractive cutoffs for spectacle provision with regards to their impact on visual improvement and spectacle compliance.
Design: Prospective study of visual improvement and spectacle compliance.
Participants: South African school children aged 6–19 years receiving free spectacles in a programme supported by Helen Keller International.
Methods: Refractive error, age, gender, urban versus rural residence, presenting and best-corrected vision were recorded for participants. Spectacle wear was observed directly at an unannounced follow-up examination 4–11 months after initial provision of spectacles. The association between five proposed refractive cutoff protocols and visual improvement and spectacle compliance were examined in separate multivariate models.
Main outcomes: Refractive cutoffs for spectacle distribution which would effectively identify children with improved vision, and those more likely to comply with spectacle wear.
Results: Among 8520 children screened, 810 (9.5%) received spectacles, of whom 636 (79%) were aged 10–14 years, 530 (65%) were girls, 324 (40%) had vision improvement ⩾3 lines, and 483 (60%) were examined 6.4±1.5 (range 4.6 to 10.9) months after spectacle dispensing. Among examined children, 149 (31%) were wearing or carrying their glasses. Children meeting cutoffs ⩽−0.75D of myopia, ⩾+1.00 D of hyperopia and ⩾+0.75 D of astigmatism had significantly greater improvement in vision than children failing to meet these criteria, when adjusting for age, gender and urban versus rural residence. None of the proposed refractive protocols discriminated between children wearing and not wearing spectacles. Presenting vision and improvement in vision were unassociated with subsequent spectacle wear, but girls (p⩽0.0006 for all models) were more likely to be wearing glasses than were boys.
Conclusions: To the best of our knowledge, this is the first suggested refractive cutoff for glasses dispensing validated with respect to key programme outcomes. The lack of association between spectacle retention and either refractive error or vision may have been due to the relatively modest degree of refractive error in this African population.
Statistics from Altmetric.com
Refractive error has been reported as the most common cause of visual impairment in a variety of school-aged populations, including children of African,1 South2 and East Asian3 and Latin American4 origin. As a result of the growing evidence that refractive error is an important cause of correctable vision loss among children, the number of school-based spectacle distribution programmes has increased dramatically in recent years. Still, little is known about children receiving spectacles in these programmes: their degree of refractive error, baseline vision, increase in vision resulting from refractive correction and their subsequent compliance with spectacles.5 Even less is understood about the impact various visual and refractive cutoffs for the dispensing of spectacles may have on important programme outcomes such as spectacle compliance and improvement in vision.
Refractive error programmes have tended to choose cutoffs on an ad hoc basis. In fact, little evidence exists in the literature to endorse the efficacy of particular refractive cutoffs in terms of well-defined outcomes. Even empirical guidelines are relatively difficult to come by. For example, in its Preferred Practice Patterns on Refractive Error,6 the American Academy of Ophthalmology suggests myopia and hyperopia cutoffs for refractive surgery, but not for the dispensing of spectacles. It may be argued that in the United States, given the relatively low cost and risk of spectacles compared with refractive surgery, empirical practice provides the best approach to this question. However, in the developing world, where vision resources for children are generally scarce, the decision as to which child should receive spectacles, and on what basis, takes on more practical importance.
We recorded demographic information, baseline vision, refractive error, and improvement in vision with refractive correction and observed spectacle compliance in a school-based refractive error programme in Umtata, South Africa, administered jointly by Helen Keller International (HKI) and Vision Care. We sought to determine the potential association between various refractive cutoffs and two principal programme outcomes: improvement in vision and compliance with spectacle wear.
SUBJECTS AND METHODS
Umtata, Transkei is located in the Eastern Cape Province of South Africa, and has a population of 72 205.7 The municipality has an unemployment rate of 57% and a poorly developed infrastructure: 6% of homes have telephones, 42% have electricity, and 9% have running water.8
HKI has partnered with Vision Care, a non-profit organisation in Umtata, to implement a school-based programme providing eyeglasses to students with refractive error. This programme, known as ChildSight®, began in January 2003. During the study period from May 2003 to May 2005, 8520 children between the ages of 6 and 19 years from 1 rural and 12 urban secondary schools in Umtata were screened, with 810 receiving spectacles.
Visual acuity screening was conducted by three project assistants after training by HKI staff. The screenings were carried out with an EDTRS chart in well-lit, quiet areas of the school when available, but the conditions were not standardised between schools. Visual acuity was measured separately for each eye of a subject while wearing habitual correction, if available. Students with presenting acuity ⩽6/12 in either eye, or whose vision improved subjectively with a +2.00 lens (indicating possible hyperopia), underwent non-cycloplegic retinoscopy and refraction by an optometrist. Guidelines were given to the optometrist for prescribing eyeglasses for myopia of ⩽−0.75D. There were no guidelines for hyperopia or astigmatism. The optometrist was given discretion to prescribe outside the guidelines. Children requiring further eye care were referred to Vision Care or the Nelson Mandela Academic Hospital in Umtata.
Spectacles were distributed to children at their schools at a later date, at which time they were adjusted for fit, and vision was again tested in each eye. Children failing to achieve the acuity measured at the time of initial refraction were referred to Vision Care for additional examination. Measurement of vision improvement as reported and analysed in this paper is calculated based on the difference between the presenting and best-corrected acuities as measured at the time of the initial examination.
Unannounced follow-up visits to the schools were conducted between 4 and 11 months after the students received their eyeglasses to determine compliance with spectacle wear. All schools included during the study period underwent such follow-up visits. All children receiving glasses at the time of initial examination and present in school on the day of follow-up were eligible for inclusion in the follow-up sample. HKI/Vision Care staff carried out direct inspection of each student who had received glasses to determine if spectacles were being worn. Children not wearing their eyeglasses were asked if they had the glasses at school. Demographic information including age, gender and urban or rural residence was recorded for each child from programme records.
Though the precise date of the follow-up visit was not announced in advance in order to more accurately assess typical patterns of spectacle wear, the purpose and methods of the follow-up study were explained to the school authorities and the School Governing Boards to give parental notification prior to data collection. This method of obtaining consent and all study procedures were reviewed by the Institutional Review Board of the Johns Hopkins University School of Medicine and the Eastern Cape Department of Education. The study was carried out in compliance with the tenets of the Declaration of Helsinki.
Univariate comparisons were made between various demographic and visual characteristics and the two principle study outcomes of interest, spectacle wear (or possession on the subject’s person) at the time of follow-up examination and improvement in vision after refraction, measured in lines, in the eye with worse presenting vision. Separate multivariate regression models (SAS, Inc., Cary, NC) were then constructed with spectacle wear/possession (logistic regression) and lines of vision improvement (linear model, vision treated as a continuous variable) as the outcome variables, and including age, gender, time since distribution of spectacles (continuous variable) and urban versus rural residence as independent variables The effect of different refractive cutoffs for the distribution of spectacles was then examined in these models to determine at what level of refractive error children meeting the stipulated criteria would differ significantly from children not meeting the criteria, with respect to spectacle wear/possession or improvement in vision. These protocols are as follows:
Current Protocol (“Protocol 0,” the actual guideline of the current programme):
(sphere ⩽−0.75 or sphere >0) OR
(sphere >−0.75 and sphere ⩽0) and cylinder <0)
Proposed Protocol no. 1:
(sphere ⩽−0.75 or sphere >0) OR
(sphere >−0.75 and sphere ⩽0) and cylinder ⩽−0.75)
Proposed Protocol no. 2:
(sphere ⩽−0.75 or sphere ⩾+1.00) OR
(sphere >−0.75 and sphere <+1.00) and cylinder ⩽−0.75)
Proposed Protocol no. 3:
(sphere ⩽−1.25 or sphere ⩾+1.00) OR
(sphere >−1.25 and sphere <+1.00) and cylinder ⩽−0.75)
Proposed Protocol no. 4:
(sphere ⩽−0.75 or sphere ⩾+1.00) OR
(sphere >−0.75 and sphere <+1.00) and cylinder ⩽−1.00)
We have used vision in the worse-seeing eye in all analyses in this study in order to mimic the standard for spectacle-dispensing utilised in the programme: if children had poor vision or refractive error in either eye, spectacles were dispensed (that is the standard for dispensing spectacles was the vision on the worse eye). All analyses were also carried out with the better-presenting vision (data not shown), with identical results.
A total of 810 children (9.5% of 8520 children screened and 38.2% of 2119 children refracted) received spectacles in the programme (fig 1). Among children receiving spectacles, 636 (79%) were between the ages of 10 and 14 years, and 530 (65%) were girls. Girls made up 52% (4436/8520) of children screened and 60% (1278/2119) of those refracted. The majority (722/810 = 89%) of children receiving spectacles in this programme were urban residents, as were the majority screened (7668/8520 = 90%) and refracted (1928/2119 = 91%). Among children given glasses, 40% (324/810) improved by 3 or more lines in the eye with worse-presenting vision (table 1). All but four children receiving spectacles had not previously worn glasses.
A number of children received glasses at the discretion of the optometrist outside the nominal guidelines of the programme (fig 1). Among children given glasses, 26% (209/810) had presenting vision of >20/30 in the worse-seeing eye (outside the vision cutoff guidelines), 9.5% (77/810) had myopia >−0.75 D and no hyperopia or astigmatism (outside the refractive cutoff protocol), and 6.2% (50/810) received spectacles outside both guidelines (table 1).
Among children receiving glasses, 60% (483/810) were present in school on the day of an unannounced follow-up examination, and of these, 31% (149/483) were wearing their spectacles (67/483 = 13.9%) or had them at school (table 2). The follow-up visits occurred at a mean 6.4±1.5 (range 4.6 to 10.9) months after dispensing of the glasses. In univariate analyses, older children were significantly (p = 0.0005) less likely to be present at school on follow-up, and those with worse-presenting vision more likely (p = 0.02) to be present. (table 2) Girls (p = 0.0005), urban children (p = 0.02) and those with worse corrected (p = 0.005) but not worse-presenting (p = 0.62) vision were more likely to be wearing/carrying their spectacles (table 2).
Separate multivariate models were created for the principle study outcomes: lines of improvement in vision after refractive correction (treated as a continuous variable) and spectacle wear/possession at the time of follow-up examination. The eye with worse-presenting vision was used in all models. All analyses were also carried out with the better-seeing eye, with identical results (data not shown).
Various refractive error cutoff protocols for the dispensing of spectacles were defined as outlined in the Subjects and methods. In models predicting improvement in lines of vision, the actual programme protocol and Protocol no. 1 (including an astigmatism cutoff of 0.75 D) were unassociated with improved vision. Children meeting cutoffs ⩽−0.75D of myopia, ⩾+1.00 D of hyperopia and ⩾+0.75 D of astigmatism (Protocols 2, 3 and 4) had a significantly greater improvement in vision than children failing to meet these criteria, when adjusting for age, gender and urban versus rural residence (table 3). Less stringent criteria failed to discriminate between children with and without improved vision. Children from urban schools had a significantly greater improvement in vision than did rural children (table 3).
In models predicting spectacle wear, girls (p⩽0.0006 in all models) and students whose follow-up examinations occurred sooner after receiving spectacles (p<0.02 in all models) were more likely to be wearing their glasses (table 4). Among the refractive protocols, none successfully discriminated between children wearing and not wearing spectacles at follow-up (table 4).
Although 75% of children receiving spectacles in this programme improved by ⩾1 line of vision, and over 200 children improved by 3 or more lines, roughly one-quarter of children received spectacles despite relatively mild deficits in refractive error (table 1). To put this number in perspective, Mitchell has reported that 38.3% of spectacle-wearing children examined on a population basis in Australia did not have significant refractive error in either eye by their study criteria.9 The goal of the current study was to establish cutoffs for the provision of spectacles that were validated with regards to important programme outcomes. The presence of children with milder degrees of visual deficit and refractive error in our sample allowed us to study how subjects meeting or failing to meet a variety of proposed refractive cutoffs differed with regards to improvement in vision and spectacle retention.
Children meeting cutoffs ⩽−0.75D of myopia, ⩾+1.00 D of hyperopia and ⩾+0.75 D of astigmatism (Protocols 2–4) had a significantly greater improvement in vision than children failing to meet these criteria, when adjusting for age, gender and urban versus rural residence. Examination of the various protocols suggests that, at least in this population, excluding subjects with modest amounts of hyperopia (<+1.00 D) from receiving glasses is more effective in discriminating between children who will and will not have improved vision than is excluding subjects with low (less than 0.75 D) astigmatism. With regards to the cutoff of −0.75D of myopia, it should be noted that we have also reported very low rates of spectacle compliance below this level, 2% (1/51) for children receiving −0.5D spectacles in Oaxaca, Mexico.5
Rates of spectacle retention in this study were modest, in keeping with findings elsewhere.5 10 Contrary to our findings among similarly aged children in Mexico, however, higher degrees of refractive error were not predictive of better spectacle compliance in this South African population. Although children with myopia <−1.25 D were 1.6× as likely to be wearing their spectacles as children without such myopia, the difference was not statistically significant. This may be due to the low prevalence of refractive errors in this population as compared with the Mexican cohort. Only 70 children (9%; table 1) in this group of children provided with spectacles had myopia ⩽−1.25D, as against the figure of 19.9% in Mexico.5 These results are consistent with other reports of relatively low refractive error prevalence among African school-age children.1 11 12
Presenting vision in the worse eye (and in the better eye in separate analyses: data not shown) and improvement in vision were likewise unassociated with glasses wear at follow-up in this cohort (table 3). It would appear that, at least in urban South Africa, strategies other than adjustment of visual and refractive criteria for spectacle provision may be needed to improve compliance with glasses. Chief among these may be programmes targeting boys, who were significantly less likely than girls to be wearing their glasses at the time of an unannounced follow-up examination. In contrast to our results from Mexico,5 urban children in this cohort had a better compliance with spectacle wear in the univariate analysis, though the proportion of rural children (11%) in the sample was small, and the apparent difference did not persist in the multivariate model. Reasons for the greater improvement in vision among urban as opposed to rural children are not clear. The amounts of refractive error did not differ between these groups; it is possible that there was a higher prevalence of non-refractive ocular disease among the rural children, though we do not have examination data to support this hypothesis.
Given our recent finding that the correction by refraction of even modest decrements in visual acuity (in the 6/9 range or above) are associated with a significant improvement in self-reported visual function,13 we suggest the lowest cutoff for spectacle provision that adequately discriminates between children who do and do not show significant vision improvement. This would be Protocol no. 2, with cutoffs of ⩾+1.00D of hyperopia, ⩾0.75D of cylinder and ⩽−0.75D of myopia.
Specific recommendations for cutoffs in the correction of refractive error by spectacles, whether supported by data or not, are difficult to come by. The American Academy of Ophthalmology Preferred Practice Patterns for Refractive Errors6 references specific FDA guidelines for the surgical correction of myopia (generally ⩽−0.5D), hyperopia (generally >+0.25 to +0.5 D) and astigmatism (generally >0.25D) but does not give specific recommendations for spectacle provision. It is noted that “patients with low refractive errors may not require corrections,”6 but no cutoffs are suggested. ICD-9CM, the International Classifications of Diseases, Clinical Modification, refers to, but does not define, myopia (367.1), hyperopia (367.0) and astigmatism (367.2).14
In seeking to define refractive error clinically, population-based epidemiologic studies have most commonly agreed on <−0.5 D for myopia,15–18 and >+0.5D for hyperopia.15 17 18 Astigmatism, when mentioned, has been defined as ⩾+0.75D15 and >+0.50D.18 A recent meta-analysis of refractive error prevalence among European-derived populations defined myopia as ⩽−1.00 D and hyperopia as ⩾+3.00 D.19
One objective factor which sets a lower bound on clinically useful refractive correction is accuracy in measurement of refractive error. It has been reported that the reproducibility of subjective refraction is in the range of 0.5D for myopia, hyperopia and astigmatism,20 21 with a 95% confidence interval reported as 0.6D.22
Small differences in refractive cutoffs may have a significant impact on programme parameters. A change in the refractive cutoff for myopia from ⩽−0.75D to ⩽−1.25D in the current programme would have reduced by 75% (210/280) the number of children receiving spectacles for myopia. An increase in the cutoff for astigmatism from ⩾0.5D to ⩾1.0D in the Mexican programme on which we have previously reported5 would have reduced astigmatic prescriptions by 51%. These differences in the number of spectacles distributed have a potentially very significant impact on the costs of refractive error programmes in the developing world, particularly when spectacles are paid for out of clinical revenues, or purchased locally and in modest amounts, which will add significantly to the unit price.
The results of this study must be taken within the context of its limitations. Our findings, particularly that refractive cutoffs were not significant determinants of spectacle use, may be applicable to the African setting only, where the prevalence of higher refractive error is relatively low. The observation that cutoffs of ⩽−0.75D of myopia, ⩾+1.00 D of hyperopia and ⩾+0.75 D of astigmatism successfully identified children with higher levels of vision improvement must also be put in context. It is possible that programmes in areas where higher refractive errors are more common may choose to utilise more stringent refractive cutoffs in order to improve spectacle compliance, as we have reported in Oaxaca, Mexico.5 It is important to note that we did not test for the impact of different vision cutoffs on visual improvement, for the reason that such cutoffs are associated almost by definition with degree of improvement in vision (a child with presenting vision of 6/7.5 cannot by definition achieve 2 lines of vision improvement on a scale where the “best” vision is 6/6). The lack of such standards in our recommendations is not meant to suggest that visual cutoffs for spectacle dispensing are not useful.
Among children receiving spectacles in the programme, only 60% (483/810) were present at school on the day of follow-up and thus examined. The fact that children who presented for examination differed from those who did not (they were younger and more likely to be from rural schools) means that that these results can be applied to the population of children participating in the programme only with care. Additionally, the true rates of spectacle wear/possession among programme participants can only be estimated from these data, as patterns of wear are not known among the 40% of non-presenters.
Non-cycloplegic refraction was used in this context, which may have led to an underestimation of hyperopia in some of these subjects (though it should be noted that only 10% of the subjects were in the age range of 6–9 years), where this problem might be expected to be of greater practical significance.
Nonetheless, the current study represents the first of which we are aware in the developing world to report refractive error cutoffs for spectacle provision which are validated with respect to important programme outcomes such as improvement in vision.
Funding: This research was supported by funding from USAID (HFP-A-00-01-00027-00) provided to Helen Keller International.
Competing interests: The authors have no financial interest in the devices and techniques reported in this manuscript.
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.