Screening for amblyopia risk factors in pre-verbal children using the Plusoptix photoscreener: a cross-sectional population-based study
- Ali Akbar Saber Moghaddam1,
- Abbas Kargozar1,
- Mehran Zarei-Ghanavati1,
- Marzieh Najjaran2,
- Vahideh Nozari2,
- Mohammad Taghi Shakeri3
- 1Eye Research Center of the Mashhad University of Medical Sciences, Khatam-al-Anbia Eye Hospital, Mashhad, Iran
- 2Khatam-al-Anbia Eye Hospital, The Mashhad University of Medical Sciences, Mashhad, Iran
- 3The Mashhad University of Medical Sciences, Mashhad, Iran
- Correspondence to Mehran Zarei-Ghanavati, Mashhad University of Medical Sciences (MUMS), Eye Research Center and Department of Ophthalmology, Khatam-al-Anbia Eye Center, Abotaleb Blvd., Mashhad, Iran;
- Accepted 28 February 2011
- Published Online First 21 April 2011
Aims To evaluate the accuracy and applicability of Plusoptix photoscreener in screening pre-verbal children at risk of amblyopia.
Methods In this cross-sectional population-based study 996 children aged 6–36 months were screened with the Plusoptix. The children underwent complete examinations that included a manual cycloplegic retinoscopy, slit lamp examination, cover test, the Hirschberg test and an assessment of fixation pattern. In addition, the prevalence of amblyogenic risk factors was estimated, the Plusopix measurements were compared with cycloplegic retinoscopy, and the referral criteria were modified to improve the accuracy of the device.
Results Amblyogenic hyperopia >+3.50 D, myopia >−3.00 D, astigmatism >1.50 D in the orthogonal meridian or >1.00 D in the oblique meridian, or anisometropia >1.50 D was detected in 4.7%, 0.0%, 12.3% and 0.5% of the samples, respectively. The average difference between retinoscopy and photorefraction for spherical equivalence was −0.16±1.0 D (p<0.05). Testability was 98.1% for the Plusoptix. The Plusoptix vision screener referred 47.8% of these children for sensitivity, specificity, positive predictive value and false-negative rates of 100%, 38.7%, 19% and 100%, respectively. These same measurements with the modified referral criteria (hyperopia ≥2.00 D in children under 12 months and astigmatism ≥1.25 D) were 93.6%, 51.2%, 21.7% and 98.2%.
Conclusion The Plusoptix is a useful objective screening instrument, but still has low specificity for detecting amblyopia risk factors in the paediatric population.
- public health
- diagnostic tests/investigation
- child health (paediatrics)
Amblyopia is the most common cause of decreased vision in children. The prevalence of amblyopia is estimated to be about 2%–5%.1 Early diagnosis through visual screening, referrals for complementary examinations and early interventions may lead to better visual outcomes2 and national screening programmes have decreased the burden of this disease in some countries.3
Plusoptix is a digital photoscreener that takes a digital video of the patient using infrared beams that are immediately analysed by the software installed on a linked laptop. This instrument is remarkable because it is portable and easy to use in examining younger children. Previous studies have shown promising results of the Plusoptix in detecting amblyopia risk factors,4–6 but due to insufficient information about its applicability in children less than 3 years of age, in this study the Plusoptix was used to screen for amblyopia risk factors in children between 6 and 36 months of age.
Materials and methods
In this cross-sectional study, sampling took place in two steps in children aged between 6 months and 3 years who were referred to Mashhad triple hygienic-therapeutic centres (three departments covering the population of Mashhad, which provide the first level of hygiene care) located in various parts of the city. Sample size was determined in each centre appropriate to the population that they covered. The sample was selected from these centres by simply admitting any children referred to these centres for vaccinations. The total sample size tested by the Plusoptix was 996 children.
The children were examined with the Plusoptix S08 photoscreener (Plusoptix GmbH, Nuremberg, Germany). This was done by two trained optometrists. Referring criteria were chosen based on a previous study by Matta et al (hyperopia ≥+3.00 D in children aged 6–12 months, +1.25 D in children aged 12–36 months; myopia ≥−2.00 D; astigmatism ≥1.00 D; and anisometropia ≥1.25 D).4 Children with one of these criteria were referred to the Khatam-al-Anbia Hospital for further examination. In addition, if a child was not cooperative, the examination was repeated up to three times in 1 day and in the case of failure, the child was referred. Children whose parents consented and were defined as healthy by the instrument also underwent the full examination process.
These examinations included a slit lamp examination, manual cycloplegic retinoscopy, cover test, binocular fixation pattern assessment (central, steady and maintained (CSM)) and the Hirschberg test. Accommodation inhibition was induced by twice applying cyclopentolate 1% drops (5 min apart) and a retinoscopy was performed after 45 min. Diagnostic criteria of amblyopia risk factors in retinoscopy were chosen based on the American Academy of Pediatrics and the American Association of Pediatric Ophthalmology and Strabismus (AAPOS) guidelines.7 These criteria include hyperopia >3.5 D, myopia >3 D, anisometria >1.5 D and astigmatism >1.5 D. If astigmatism axis was oblique (more than 10° from 90° or 180°) values >1 D were considered positive.
Specificity and sensitivity of this instrument were defined in comparison with retinoscopy. In order to assess the best criterion for referring children by the Plusoptix instrument examination, the receiver operating characteristic curve (ROC curve) was used. This curve was used to assess a measurement for hyperopia in two groups, one aged <12 months and the other one aged between 12 and 36 months, and astigmatism in the group aged 6–36 months. Considering the few number of children with myopia and anisometropia, the best referring measurements for these factors were not assessed.
To define the repeatability of the measurements, the interclass correlation test was used. Sphere and cylinder amounts that were obtained by the Plusoptix and retinoscopy were compared by the paired t test and analyses were done by SPSS software version 16 (SPSS Inc., Chicago, Illinois, USA). A p value <0.05 was considered to be statistically significant.
In the children studied (996 cases) the average age was 16.95±8.59 months. They were at least 5.5 months old and at most 36.8 months old; 52.5% of this population was male and 47.5% was female. The mean sphere of the right eye recorded by the Plusoptix was +1.38±0.95 D, the mean cylinder was −0.85 ±0.71 D and the mean spherical equivalent was +0.96±0.82 D. Of the tested children 47.8% were referred for complementary examinations after being examined with the Plusoptix. Reasons for referral based on the Matta criteria were astigmatism (35.4%), hyperopia (20.3%), anisometropia (1.9%) and myopia (0.3%) (some cases had more than one risk factor). We were not successful in assessing 17 children (1.9%) with the Plusoptix because they were uncooperative or had miotic pupils.
A total of 382 children received complementary examinations in Khatam-al-Anbia Hospital. Among these, 260 of children (68.1%) were referred by the instrument (having risk factors or being uncooperative with the test) and 122 (32.0%) were diagnosed as healthy by the instrument but their parents volunteered for complementary examinations. Two hundred and thirty-three failed children (32.1%) declined to participate in complementary examinations in Khatam-al-Anbia Hospital. The average age of this group was 16.36±8.5 months (47.9% were female and 52.1% were male). In the studied population, the spherical equivalent, sphere and cylinder of the right eye obtained by retinoscopy were +1.30±0.99 D, +1.64±1.06 D and −0.67±0.72 D, respectively. Amblyopia risk factors for one eye or both eyes were detected by the complementary examinations in 15.2% of children, while it was found that 12.3% of the children had astigmatism, 4.7% hyperopia, 0.5% anisometropia and 0% myopia. Alternative esotropia of approximately 15 prism dioptres was detected for one patient for whom the instrument had given the diagnosis of high astigmatism. In the Plusoptix diagram for this child, alignment dots were between 10° and 20° circles (figure 1).
Out of 17 patients in whom the instrument was not capable of performing the test, eight patients (included in 382 analysed cases) came for complementary examinations and only one of the eight patients actually had high astigmatism defined as an amblyopia risk factor.
One person performed measurements three times in 99 children, who had an average age of 15.09±8.09 (range 6 to 36) months. In order to assess the repeatability of the instrument, the interclass correlation test was used. Correlation for measuring the sphere was 0.904 and for the cylinder it was 0.972, which showed the high repeatability of the instrument for measuring the sphere and cylinder.
The Plusoptix instrument had a sensitivity of 100% and specificity of 38.7% in the diagnosis of all amblyopia risk factors. Negative predictive value, positive predictive value and false positive were 100%, 19% and 56.5%, respectively (table 1).
In children aged <12 months, the instrument had a sensitivity of 25% and specificity of 97.6% in the diagnosis of hyperopia risk factor. Negative predictive value and positive predictive values were 96.4% and 33.3%, respectively. Considering the ROC curve (with the area under the curve being 0.796), sensitivity of 71.5% and specificity of 86.1% was obtained by using the hyperopia referral level of 2.00 D (figure 2). In children aged >12 months, the Plusoptix was found to have a sensitivity of 80.0%, specificity of 59.2%, negative predictive value of 99.1% and positive predictive value of 4.8% in the diagnosis of amblyogenic hyperopia. The ROC curve (with an area of 0.738 under the cover) showed that specificity could not be increased without decreasing the sensitivity, and the criterion of 1.25 D was satisfactory for hyperopia.
The Plusoptix instrument had a sensitivity of 97.4% and specificity of 52.4% in the diagnosis of amblyopia inducing astigmatism. The negative predictive value and positive predictive value was 99.4% and 19.2%, respectively. Considering the ROC curve, by increasing the referral measure of astigmatism to 1.25 D and decreasing the sensitivity to 84.6%, specificity was increased to 73.4% (figure 3).
According to table 2, by changing the referring criterion for astigmatism to 1.25 D and for hyperopia to 2.0 D in children <12 months, we reached a sensitivity of 93.6% and specificity of 51.2%. The negative predictive value and positive predictive value was 98.2% and 21.7%, respectively. By applying these criteria three patients (0.7%) were false negative with astigmatism of 1.75 D. False positive was 46.6%.
The paired t test was done in order to compare the sphere, cylinder and spherical equivalent measures of the Plusoptix instrument with the retinoscopy. The average differences of the sphere, cylinder and spherical equivalent were +0.05±1.05, −0.43±0.58 and −0.16±1.0 D, respectively. Therefore, the Plusoptix found more hyperopia/less myopia than the retinoscopy for the sphere, and more myopia/less hyperopia for the cylinder and spherical equivalent. Although the difference between the sphere measurements of the Plusoptix instrument and retinoscopy were not statistically significant (p=0.43), the cylinder and spherical equivalent were statistically different (p=0.01). The difference in the sphere, cylinder and spherical equivalent measured by the Plusoptix and retinoscopy was ±1.0 D in 27.6%, 13.9% and 26%, respectively (figure 4).
This is the first population-based study for assessing the results of the Plusoptix in amblyopia screening in children aged between 6 and 36 months. Our experience with this instrument reveals its simple usage for screening in the community. In addition, data gathering was easily done with the Plusoptix's database. Bearing in mind that this study is unusual in being population-based and in the age range of the children entered into this study, it is complementary to previous studies.
In our study, the mean difference in spherical equivalent measured by the Plusoptix and retinoscopy was 0.16 D, but Erdurmus et al showed that the spherical equivalent of the instrument was 0.63 D more myopic and some of the children had an accommodation of up to 3 D.8 Schimitzek and Lagrèze noted a 0.73 D shift to myopia due to uncontrolled accommodation.9 Dahlmann-Noor et al reported a myopic shift of 1.9 D.5 These findings show the possibility of uncontrolled accommodation in patients who were examined by the Plusoptix, which according to the findings of Schimitzek and Haase and of Schaeffel et al cannot be relieved by using cycloplegic drops or 3 D glasses.10 11 The difference between our findings and these studies is probably because of the difference in age. These studies are usually done in children >3 years of age. In the study of Erdurmus et al, the average age was about 7.5 years while in our study it was 1.5 years.8 Perhaps in younger children the camera of the instrument does not act as an accommodation stimulator.
In this study, the instrument showed high repeatability. In the study of Dahlmann-Noor et al, the repeatability coefficient was about 0.6 D, although only 35% of retinoscopy and Plusoptix results were within ±1.0 D of each other (74% in our study).5 This shows the good repeatability of the instrument. In addition, the testability with this instrument was >98%, which is high considering the young age of the children. This measure was 100% in another study by Dahlmann-Noor et al.6
In this study the Plusoptix instrument had a sensitivity of 100% and specificity of 38.7% in diagnosing amblyopia risk factors. The false predictive value and positive predictive values were 100% and 19%, respectively. We used the criteria suggested in the study of Matta et al,4 which showed a sensitivity of 98% and specificity of 69%. The same researchers reported a sensitivity of 98.9% and specificity of 96.1% in their next study using similar criteria.12 In the population studied, 65% of patients were affected by amblyopia or had amblyopia risk factors. In both studies done by Matta et al, patients were chosen from a clinic and the prevalence of children with amblyopia was high. In our study, children were chosen from the city population and the prevalence of amblyopia risk factors was 15.2%. Therefore, we reached lower specificity than previous studies. Dahlmann-Noor et al compared the results from the Plusoptix instrument with orthoptic assessments in a study done in children aged 4–7 years.6 As the referral criteria for the Plusoptix, they considered hyperopia >3.0 D, myopia >−1.0 D, astigmatism >1.5 D and anisometropia >1.0 D. They reported a sensitivity of 44% and specificity of 100% for the Plusoptix, which shows unsuitable criteria. Sensitivity and specificity of the Plusoptix for diagnosing the amblyopia risk factors in the study of Clausen and Arnold were 67% and 94%, respectively.13 They suggested the referral criteria for the Plusoptix as hyperopia >2.5 D, myopia >−2.0 D, astigmatism >1.5 D and anisometropia >1.0 D. Table 3 shows a comparison of the sensitivity and specificity for Plusoptix obtained from previous studies with ours.
In our study sensitivity was reduced when astigmatism was defined as >1.25 D. By defining the astigmatism referral criterion as 1.5 D, sensitivity was reduced to <90% but the specificity did not rise significantly. By applying Matta's referring criteria we had a sensitivity of 100%. Thus, no case of amblyopia risk factor was missed. On the other hand, the instrument referred nearly half of the children, which is high for a screening device. The low specificity of the Plusoptix increases the cost of mass screening and one solution is the improvement of referral criteria. Referred children could be examined by a more specific screening device, that is by designing a two-step screening assessment. It can be done for example by using Sure Sight Vision Screener with specificity set to 90%.14 Future studies will reveal the efficacy of such screening programmes. In comparison with the Vision In Preschoolers Study, when specificity was set at 90%, sensitivities of non-cycloplegic retinoscopy were nearly 60% in the Retinomax Autorefractor, Sure Sight Vision Screener and Lea Symbols test.14 The MTI photoscreener had a sensitivity and specificity of 65% and 87%, respectively, in screening for amblyopia in pre-verbal children.15
A limitation of this study was that not all of the children who were examined by the Plusoptix underwent complementary examinations. Among these children, only one case of strabismus was found and no myopic patient was detected. Thus applicability of the instrument in these cases cannot be assessed.
In conclusion, the Plusoptix instrument benefits from a good repeatability and applicability in amblyopia screening in children aged from 6 to 36 months. Measuring the refraction with this instrument is to a great extent similar to retinoscopy. Performing screening in the general population using the previous suggested criteria will provide low specificity but changing these criteria can raise the accuracy of this instrument.
We thank all of the children and parents for participating in the study. We also thank the Welfare Organization of Khorasan Razavi for supporting this project. Gratitude is also extended to Mr Jalil Rahimi, Khatam-al-Anbia Eye Hospital, for performing the retinoscopy.
Funding The study was funded by Mashhad University of Medical Sciences and Welfare Organization of Khorasan Razavi.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Mashhad University of Medical Sciences.
Provenance and peer review Not commissioned; externally peer reviewed.