Refractive errors (e.g., myopia, hyperopia, and astigmatism) are important vision disorders requiring screening in children and adults. Although screening using a visual acuity chart is the most practical method to detect refractive errors, the methods and thresholds of screening have not been universally accepted. Early detection of refractive errors in children allows timely intervention in the form of spectacle correction.
For the purpose of screening elementary schoolchildren, two types of visual acuity charts have been most commonly used. The first type, the non-logarithm of the minimum angle of resolution (MAR) type of visual acuity chart, uses lines of English alphabets, with no universally accepted number of letters per line or standardized number of lines per chart, and has acuity values recorded as a Snellen notation (a fraction) or as a decimal. The second type, from the Early Treatment Diabetic Retinopathy Study (ETDRS), is a letter-counting type of visual acuity chart, which not only has a standardized number of lines and letters per line but also is able to discriminate finer levels of visual acuity and document minimum resolution acuity in a logarithmic scale (logMAR), which facilitates algebraic operations for compiling statistics.
Despite the shortcomings of using the non-logMAR visual acuity chart and the advantages of the second type of visual acuity chart, screening traditionally has been performed in many studies using simplified visual acuity charts based on Snellen fractions. Various studies using a “Snellen” chart for screening in children have been reported in the U.S., Western Australia, Denmark, Oman, and England using an equivalent of “6/12 or worse” criterion without previous justification of this threshold. For methodological reasons discussed elsewhere, these previous studies were also unable to determine the actual sensitivity and specificity of screening. In contrast, in many scientific studies, especially clinical trials, in which visual acuity and its changes were important endpoints, visual acuity has been measured using an ETDRS type chart and documented in logMAR format. The optimal logMAR visual acuity threshold for screening referral recently has been reported.
In Singapore, a multilingual society with a high rate of English literacy, simple 7-line visual acuity charts with English alphabets have been used for many years to screen for eye problems in schools. Children with visual acuity of “6/12 or worse” are referred for further assessment. The accuracy of screening using the non-logMAR type chart compared with the ETDRS chart has not yet been evaluated in a large study involving schoolchildren.
The objective of this study was to compare the sensitivity and specificity of a method of screening using a simple 7-line acuity chart with a more exacting method using the ETDRS chart for the purpose of screening for refractive errors in Singapore schoolchildren. As a secondary objective, the study aimed to estimate the best cutoff values for the detection of refractive errors.
DISCUSSION
The two main findings in this study were the optimal referral thresholds for the two methods of measurement of visual acuity and the greater relative accuracy of the logMAR visual acuity measurement by optometrists to correctly predict cases of myopia and any refractive errors in children.
Bearing in mind that different types of professionals were used for the visual acuity measurements, this study showed that the optimal threshold level for using the visual acuity chart to screen for myopia is similar between the ETDRS and the simplified screening charts. In the case of the prediction of any refractive errors, the optimal threshold for referring cases on the logMAR visual acuity appears to be lower than the case for the simplified acuity charts. Nevertheless, it would appear from the data that the intuitive threshold of 6/12 used in screening seems to be the optimal level.
The determination of the optimal threshold in this study assumes the sensitivity and specificity of screening to be equally important. In populations with different prevalences of myopia and refractive errors, this assumption may not be valid. In populations with low prevalences of myopia, for example, it may be preferable to adopt a more specific test that may result in fewer false-positive referrals.
The ROC curve for the ETDRS chart was higher than the simplified visual acuity screening chart. This implied that for any threshold on the simplified chart as used by nurses, there would exist a superior threshold using the logMAR chart by optometrists in terms of sensitivity and specificity for the detection of refractive errors. This superior accuracy of the logMAR has never been demonstrated in any previous study. For given levels of sensitivity, the specificity values of the ETDRS method were superior. At the 91% sensitivity level, the specificity is about 88% for the ETDRS and 83.5% for the method using the simplified screening chart. Although the superiority of the ETDRS method is clear, economic considerations will dictate whether this level of benefit warrants a switch in real screening scenarios.
Camparini et al. have shown that a fast ETDRS threshold testing method to measure logMAR visual acuity is valid compared with the full threshold testing method. In this article, we have taken this one step further to show that in a school population, which is a common target for screening for refractive errors, the logMAR visual acuity test, using this fast threshold algorithm, performed superiorly by optometrists compared with the current standard of visual acuity screening as performed by school nurses.
What are the possible reasons for this relative superiority in the case of screening with logMAR visual acuity measurement by optometrists? Besides the issues of standardized letter sizes and spacing, this study used the line-by-line scoring with the simplified screening visual acuity chart, a common practice with this type of visual acuity chart. In contrast, letter-by-letter scoring was performed using the ETDRS charts. Letter-by-letter scoring has been shown to result in improved test-retest variability than is permitted by line-by-line scoring. In this particular study, the testing of the children was performed by different testers for the two different types of visual acuity charts. It may be that optometrists could perform the testing procedure more rigorously than nurses. In the case of the simplified screening visual acuity charts, there is a possibility of some children artificially obtaining better than real visual acuity in the left eyes from the effect of memorizing the letters while reading with the right eyes.
The mean difference between the logMAR visual acuity compared with the simplified screening acuity was significantly different from zero. This suggests that visual acuities measured on the simplified charts have a tendency to be worse. A systematic difference exists despite the presence of a good correlation, which measures degree of association but not agreement.
The strengths of this study include a large sample size drawn from a population, uniformity of assessment, and objectivity of autorefraction. The use of cycloplegia excluded pseudomyopia or accommodative spasm.
Economic and logistic considerations, important in screening tests, have not been considered in this article. Initial impression of the testing indicated that the time required for measuring visual acuities with the logMAR method was slightly longer than that for the simplified screening acuity charts (unpublished data).
The study sample was not randomly selected from the school population of Singapore. The method of sampling was an issue in this study. The sensitivity and specificity profile may change with differing disease prevalences in different schools. Berkson’s fallacy dictates that in a sample obtained from high-risk and low-risk populations, a biased sensitivity estimate is obtained in the high-risk population and a biased specificity estimate is obtained in the low-risk population. These limitations of the study make it more difficult to generalize our findings to other populations.
A further limitation of the study is the time interval between the two methods of visual acuity assessment; in a period of up to 4 months, it is possible that some of the subjects may have had progression of their refractive errors (in particular, myopia). The direction of bias introduced is uncertain. To evaluate the possible effect of bias, the adjusted scores ([kappa]) for sensitivity and specificity also were calculated. The result, once again, favors the logMAR visual acuity measurement.
For now, we conclude that the advantage of using the ETDRS method for screening for refractive errors is at least of statistical significance. This advantage may be related to the nature of the visual acuity chart or to the different background of the screeners. Should the cost-effectiveness of screening be equivalent for the two methods of determining visual acuity, the logMAR method of screening is preferred for detection of myopia or any refractive errors in a population like Singapore, where there is a relatively high prevalence of refractive errors, particularly myopia.
No comments:
Post a Comment