Aim To compare sphere and cylinder refraction values using retinoscopy and autorefraction under cycloplegic conditions in children.
Methods This cross-sectional study was carried out using multistage cluster sampling. The target population was children aged 6–12 years in Shahroud, a northern city in Iran. Examinations included measurements of visual acuity, subjective refraction and objective refraction. Objective refraction was measured with and without cycloplegia with a retinoscope and an autorefractometer.
Results After applying the exclusion criteria, data from 5053 children were analysed. Spherical refraction results with autorefraction were significantly higher than results with retinoscopy (P<0.001). Refraction overestimation was significant in all age groups (P<0.0001). Comparison of differences in different spherical ametropia subgroups also showed a significant intermethod difference in all refractive states (P<0. 01). Overall, autorefraction tended to over plus hyperopics and under minus myopic cases compared with retinoscopy. The 95% limits of agreement for spherical values measured with the two techniques were −0.35 Diopter (D) to 0.50 D. The values of J0 and J45 vectors with autorefraction were significantly higher than those with retinoscopy (P<0.001). The 95% limits of agreement between the two methods for vectors J0 and J45 were −0.12 D to 0.15 D and −0.10 D to 0.11 D, respectively.
Conclusion Since the observed differences in spherical refraction and the cylindrical components obtained through retinoscopy and autorefraction are statistically significant, but clinically insignificant, and the two methods have a strong correlation and agreement, it can be concluded that autorefraction can be a suitable substitute for retinoscopy in children under cycloplegic conditions.
Statistics from Altmetric.com
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.
Refractive errors, as the most common eye disorder, have attracted a great amount of research.1 2 Various measurement methods have been developed to measure refractive errors, such as retinoscopy, subjective refraction, autorefraction and photorefraction. Various tools and devices have also been produced and supplied by various companies for this purpose. One of the most important points in refraction is the accommodation status during the test.3–5 In fact, in order to obtain an accurate measurement of refractive errors, accommodation must be in a relaxed condition, and there should be no accommodative response as far as possible. Naturally, any accommodation during the test will cause a negative shift in refraction, resulting in the overestimation of myopia and underestimation of hyperopia, and thus, a case who is in fact hyperopic or emmetropic may be diagnosed as myopic.6–8 On the other hand, accommodation can also cause changes in the cylindrical component of refraction, and research has shown that the astigmatism measured during refraction under dynamic conditions is different from static conditions.9–11 It is also suggested that the cylindrical component in dry refraction is different from cycloplegic readings.5 12 Therefore, under cycloplegic conditions, all refractive error components can be different.
Accurate measurement of refraction is one of the most important concerns in eye examinations. In some cases, it becomes even more important, such as reporting results of epidemiological studies of refractive errors, measurements and prescriptions in children and amblyopia treatment, precise prescriptions for best vision and minimum asthenopia especially for people who need sharp vision, such as computer users, accurate determination of refractive errors prior to refractive and cataract surgery, etc. Numerous studies have compared various methods and instruments that are used for measuring refractive errors; these include comparisons of retinoscopy results as the gold standard with various methods such as autorefraction, photorefraction and subjective refraction.13–16 On the other hand, some studies have compared the results of cycloplegic refraction with manifest refraction. According to studies, cycloplegic refraction is more accurate than dry refraction and is the preferred method when precision matters.16–19 Most studies agree that dry autorefraction results differ from that with retinoscopy and subjective refraction, and it is not very reliable, especially in children, while cycloplegic autorefraction results are reliable and accurate.16 17 20 Some studies even suggest that cycloplegic autorefraction is the most reliable method, and offers better reliability than subjective refraction and retinoscopy.21 In clinical practice, larger differences are seen between results of cycloplegic refraction obtained with retinoscopy and autorefraction in myopic cases, and it seems that autorefraction overestimates myopia even under cycloplegic conditions. Therefore, in order to resolve the contradiction in the results of previous studies, as well as to test the mentioned clinical evidence, the present study aims to compare cycloplegic refraction results between retinoscopy and autorefraction in a large sample of children.
The present cross-sectional study is part of the first phase of the Shahroud Children Cohort Study conducted in 2015 in Shahroud County, in the north of Iran. The target population was the elementary school children in urban and rural Shahroud. The entire rural population of elementary school children (n=1214) was invited to the study. In urban areas, cluster sampling was conducted. Students in urban areas were in 473 classes, and to reach a sufficient sample, 200 of the 473 classes were selected as clusters using systematic random sampling.
On participation of the children in the study and obtaining parent consents, they were all transferred to the study site free of charge, where vision tests, ophthalmic examinations and imaging tests were carried out.
In the first stage, demographic information was collected through interviews, and then visual acuity tests were done. First, uncorrected visual acuity was measured at a distance of 3 m using the Nidek CP-770 chart projector. All students had non-cycloplegic refraction with the Nidek ARK-510A autorefractokeratometer. For all students, autorefraction results were refined using the Heine Beta 200 retinoscope (HEINE Optotechnic, Hersching, Germany), and eventually, all students underwent subjective refraction.
At the end of the examinations, students were subjected to cycloplegic refraction. For this purpose, two drops of cyclopentolate 1% were instilled in the eyes at a 5 min interval, and refraction was performed half an hour after the second drop. Cycloplegic refraction was performed using autorefraction and retinoscopy, and the results were recorded separately.
Any case with a history of eye surgery, history of ocular trauma, tropia or missing refraction data was excluded from the study.
In this study, retinoscopy and autorefraction were compared by analysing data on spherical error and spherical equivalent refraction, and to compare cylinder, J0 and J45 vectors were used. The J0 vector is the horizontal and vertical components of astigmatism in each eye, and the J45 vector is the oblique component of astigmatism. These vectors were calculated using the following formulas:
Where C and α represent the magnitude and direction of cylindrical refraction, respectively. The analysis was performed using Stata and MedCalc software. In this study, mean values of sphere, spherical equivalent, J0 and J45 were compared between the two methods using paired t-test. Pearson’s correlation coefficient was used to show the correlation between the values from the two methods. To illustrate the agreement between the two methods, the Bland and Altman graphs were used with 95% limits of agreement. In the Bland and Altman graph, the horizontal axis is the mean value of sphere, spherical equivalent, J0 and J45 of the two methods and the vertical axis is the difference between the two methods in these variables. The 95% limits of agreement were calculated based on ‘mean±1.96×SD’ of the intermethod differences.
The study was conducted in accordance with the tenets of the Declaration of Helsinki. All parents signed a written informed consent and verbal consent was obtained from the children.
Of the 6624 randomly selected students, 5620 participated in this study. After applying the exclusion criteria, data of 5053 children were analysed. Due to the high correlation between the refraction results of fellow eyes, analysis was performed on results of right eyes. The mean age of the participants was 9.20±1.71 years (6–12 years) and 52.4% (n=2647) of the studied students were boys.
Results of the analyses on the differences between the refraction values obtained from retinoscopy and autorefraction are presented in tables 1–4. Results of sphere and spherical equivalent values obtained by cycloplegic retinoscopy and cycloplegic autorefraction are presented in tables 1 and 2.
As shown in tables 1 and 2, mean sphere and spherical equivalent refraction values were significantly higher with autorefraction compared with retinoscopy. Except for the last age group which was marginally significant, this difference was statistically significant in all ages. Analysis in different groups of spherical refractive error showed statistically significantly higher values of sphere and spherical equivalent with autorefraction compared with retinoscopy in all groups, except myopia >2 Diopter (D). Also, the correlation between the two methods was 0.969 for sphere and 0.970 for spherical equivalent values, and 95% limits of agreement were from −0.35 to 0.50 D for sphere and from −0.39 to 0.48 D for spherical equivalent. Table 1 summarises the limits of agreement between the two methods by age and refractive error. Figure 1 shows the Bland-Altman graphs for sphere and spherical equivalent measurements.
The two main vectors of cylinder refraction, including J0 and J45 vectors, were also compared. The results of this comparison are presented in tables 3 and 4, respectively. As shown in table 3, the J0 values were significantly higher with autorefraction than with retinoscopy. This difference was significant in all ages except those aged 12 years. A comparative study in different levels of spherical ammetropia showed that the difference in J0 values was significant only in hyperopic groups. The correlation of the two methods in determining J0 was 0.969, and the 95% limits of agreement between the two methods were from −0.12 to 0.15 D. Intermethod correlation values and limits of agreement based on age and refractive error are summarised in table 3. Figure 2 also shows Bland-Altman graphs for the two methods.
According to table 4, the values of the J45 vector derived from autorefraction were significantly higher than those with retinoscopy. This difference was also studied in different groups of spherical refractive error. The differences in J45 values were statistically significant only in the emmetropia and low hyperopia groups. The intermethod correlations and limits of agreement for J45 values are shown in table 4, and the Bland-Altman graph is illustrated in figure 2.
The present study is one of the few studies to compare cycloplegic refraction results between retinoscopy and autorefraction, and new results are being presented. One of the most important aspects of the present study is the comparison of astigmatism values in vector form. This study has been conducted on a considerably large sample of children, and given the sample size, the results can add valuable new information to existing knowledge.
The findings of this study showed that refractive values derived from retinoscopy and autorefraction are different even in cycloplegic conditions, and spherical refraction results with autorefraction tend to be over minus in myopic and emmetropic cases and over plus in hyperopic cases. Previous studies mainly investigated the difference between autorefraction and retinoscopy results under non-cycloplegic conditions, and reported that autorefraction generates over minus results compared with retinoscopy in children, and this has been attributed to the inability of the autofogging system in autorefractometers.22 23 Some previous studies have compared autorefraction results between cycloplegic and dry conditions. These studies have reported that in younger ages, autorefraction results would be over minus due to lack of adequate control of accommodation, and measurements are not reliable in children.16 20 23 The comparison in the present study is slightly different from previous studies. This study showed that autorefraction under cycloplegia generates over plus results overall. A similar study by Prabakaran et al reported that autorefraction results compared with retinoscopy, even in cycloplegic conditions, were over minus (mean difference=−0.06), which was statistically significant but clinically Insignificant.14 In fact, our results are not consistent with the study by Prabakaran et al. The discrepancy between these two studies can be attributed to several factors. First, the sample size in this study was 5054, but only 51 in theirs. Second, their study was conducted on a sample of age group 2–3 years, that is, a younger sample and a smaller age range. Other reasons can be use of different equipment, because different devices may generate different results. Due to the large sample size, the results of the present study can have preference over the previous study. One of the strengths of the present study was investigating the difference in refraction results between retinoscopy and autorefraction by type of refractive error and in different age groups, which had not been done in previous study. The significance of the difference in all studied age groups, as well as in various refractive groups (myopia, hyperopia and emmetropia) can show that the difference is not just by chance. It seems that the overestimation observed in this study cannot be attributed to accommodation for two reasons. First, results have been obtained under cycloplegic conditions, and accommodation has mainly been controlled. Second, if accommodation were the cause, we would observe an underestimation in hyperopics, but overestimation was observed even in hyperopics. Therefore, some factor other than accommodation is involved. A possible explanation is that perhaps autorefraction is more accurate than retinoscopy when performed under cycloplegia, especially in hyperopic cases. In other words, it is true that autorefraction does not provide an accurate measure of hyperopia in non-cycloplegic conditions because of the interference of accommodation, but it may provide more accurate measurements of refractive errors under cycloplegia when the role of accommodation is eliminated.
Regarding clinical applications, although the differences observed in spherical and spherical equivalent values between autorefraction and retinoscopy in children was statistically significant in this study, they can be considered clinically insignificant, because they are <0.25 D. It should be noted, however, the difference observed in the >2 D hyperopic group was about 0.25, which may be considered clinically significant. The correlation of the results obtained with these two methods under cycloplegic conditions was higher than 0.9 for all age groups and types of refractive error; this correlation is strong and significant, and on the other hand, the 95% limits of agreement for spherical refractive error values was quite a small range. This suggests that these two methods are interchangeable in the clinic when used under cycloplegic conditions. Since the difference in sphere values measured with the two devices was not clinically significant, there is no need to apply a correction factor. In fact, perhaps the most interesting and important observation in this study was that, contrary to expectations, autorefraction no longer underestimated hyperopia under cycloplegia. In fact, in cycloplegic conditions, hyperopia is overestimated with autorefraction compared with retinoscopy.
Comparative studies on astigmatism have also shown significant differences in the values by autorefraction and retinoscopy. Overall, cylinder values with autorefraction are significantly overestimated compared with retinoscopy. In a similar comparative study of cycloplegic values using retinoscopy and autorefraction, cylinder values showed a significant difference.16 There is agreement between these two studies in this regard. However, since the comparison of astigmatism was not done with vectors in that study, and different groups by refractive error were not compared, there is not much more to discuss and compare between our study and theirs. In a 2004 study by Zhao et al who tested the accuracy of refraction measurements in non-cycloplegic conditions for children aged 7–18 years, the difference between autorefraction and retinoscopy for J0 and J45 vectors was statistically significant, but clinically insignificant. In other similar studies, it has been reported that in non-cycloplegic conditions, autorefraction has adequate accuracy and reliability for measuring cylinder in children.24 25 In this study, although the difference between cylinder values using autorefraction and retinoscopy in cycloplegic conditions was statistically significant, it was clinically non-significant. In fact, the statistical significance of the difference can be attributed to the large sample size, low SEs and resulted high analysis power. In this situation, null hypotheses can be rejected even with clinically not important differences. In this study, since the correlation between cylinder results obtained with the two methods was high and the 95% limits of agreement had a narrow range, it can be concluded that, under cycloplegic conditions, autorefraction is a suitable replacement for retinoscopy for the assessment of astigmatism in children. In a study by Walline et al to compare the reliability of astigmatism measurements by three methods of autorefraction, subjective refraction and retinoscopy, the most reliable method for assessing astigmatism was cycloplegic autorefraction followed by dry autorefraction and cycloplegic subjective refraction.22 In their study, dry retinoscopy showed the least reliability.21
In conclusion, the results of the present study showed that in children under cycloplegia, autorefraction provides values comparable to retinoscopy. The sphere and cylinder values obtained in clinic with this method can be reliable, and if the child is cooperative for proper fixation (having good stable fixation to autorefractometer’s target), the results of cycloplegic autorefraction can be used directly for the prescription. An interesting point in this study was that in all refractive states, even in hyperopia, autorefraction values are overestimated compared with retinoscopy, which may be slightly significant in high hyperopia. But the amount of difference is clinically negligible. On the other hand, in epidemiological studies of refractive errors, autorefraction can be used in place of retinoscopy in children under cycloplegic conditions, which will certainly speed the process and reduce the cost of screening.
The authors would like to thank children and their parents who participated in this study.
Contributors HH and AA drafted the manuscript and contributed in preparation of the study protocol and conceptualised and conducted all statistical analyses and were the primary author of the article. AY and MK contributed in the conceptualisation of the paper and the statistical analyses and critically revised the manuscript. MHE and AF conceived and designed the study and contributed in preparation of the study protocol and contributed in the conceptualisation of the paper and the statistical analyses and critically revised the manuscript.
Funding Shahroud School Children Eye Cohort Study is funded by the Noor Ophthalmology Research Center and Shahroud University of Medical Sciences (project number: 9329).
Competing interests None declared.
Ethics approval The ethics committee of Shahroud University of Medical Sciences, Shahroud, Iran.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Any new paper about Shahroud School Children EyeCohort Study is welcome. The researchers can contact MHE by firstname.lastname@example.org.
Presented at Obtained.