Patients characteristics
One hundred patients were included between September 2016 and June 2017. The average age was 68 months. Four children were excluded from the analyzes because they could not perform the tests. Patient characteristics are presented in Table 1.
Table 1
Criteria collected | Number |
Age : (in months) | |
- Between 36 and 59 | 28% |
- Between 60 and 66 | 22% |
- Between 67 and 79 | 23% |
- Between 80 and 112 | 24% |
Sexe : | |
- Girls | 56% |
- Boys | 44% |
Number of consultations préalables | |
- First consultation | 16% |
- Between 1 and 5 | 38% |
- Between 6 and 10 | 21% |
- > 10 | 25% |
Reason for consultation | |
- Screening | 25% |
- Amblyopia or known risk factor of amblyopia | 75% |
Known amblyopia | 12% |
Strabismus | 34% |
Hypermétropia | 22% |
Astigmatism | 10% |
Myopia | 4% |
Duration of use of a smartphone or tablet at home per week | |
- < 3 hours | 62% |
- between 3 and 7 hours | 22% |
- between 7 and 14 hours | 13% |
- > 14 hours | 3% |
Main outcome (Table 2)
The average visual acuity obtained with the Rossano-Weiss test was − 0.22 logMAR or 6.2/10 for the right eye and − 0.24 logMAR or 6.1/10 for the left eye. The mean visual acuity obtained with the eMOVA test was − 0.28 logMAR or 5.9/10 for the right eye and − 0.24 logMAR or 6.1/10 for the left eye. The mean difference in visual acuity between the two groups was − 0.06 logMAR or 0.3/10 for the right eye and 0.00 logMAR or 0/10 for the left eye. The maximum visual acuity obtained was − 0.18 logMAR for both tests. The minimum visual acuity achieved was − 0.67 logMAR and − 1.2 logMAR, respectively, for the Rossano-Weiss test and the eMOVA test. The first quartile, median, third quartile and mode were − 0.18 logMAR for all groups.
Overall, the differences in visual acuity measured by the two tests are very limited. In order to test the equivalence of the two tests, a statistical analysis of concordance was conducted.
For this analysis, we studied two sets of values to verify the validity of our results: the first on the right eye and the second on the left eye of each patient.
The correlation coefficient r, of the visual acuity measurement with the Rossano-Weiss test and the eMOVA test was 0.40 (p < 0.001 and 95 [0.21–0.55]) for the right eye series and 0.43 (p < 0.001 and IC 95 [0.26–0.58]) for the left eye series. Ninety-one percent (91%) of the right eyes, whose visual acuity measured with the Rossano-Weiss test was − 0.18 logMAR, achieved the same visual acuity with the eMOVA test (respectively 95% and 0.18 logMAR for left eyes). The average difference between the two tests was − 0.06 logMAR (lower bound: -0.48, upper bound 0.36) for the right eye series and − 0.01 logMAR (lower bound: -0,40, upper bound: 0.38) for the left eye series.
Table 2
| Right eye series | Left eye series |
Average visual acuity (in logMAR) | | |
- Rossano-Weiss test | -0.22 | -0.24 |
- eMOVA test | -0.28 | -0.24 |
Average difference between the two tests | -0.06 [-0.48–0.36] | -0.01 [-0.40–0.38] |
p | 0.006 [-0.10–0.02] | 0.006 [-0.10–0.02] |
Correlation coefficient | 0.40 | 0.43 |
On the Bland and Altman analysis (Fig. 1), the concordance is very good for the highest visual acuities and from − 0.6 logMAR, the concordance becomes less good. The eMOVA test overestimated the visual acuity of the lowest acuities.
The eMOVA test overestimated the visual acuity of 0.06 logMAR for the right eye series and 0.01 logMAR for the left eye series.
Using the lower and upper limits, we find that at most the difference in values obtained between the two tests was − 0.48 logMAR for the right eye series (Fig. 2) and − 0.40 logMAR for the left eye series (Fig. 3).
The average difference between the two measurements was − 0.06 logMAR in favor of the eMOVA test using the Wilcoxon test. This difference was statistically significant (p = 0.006). The 95% confidence interval of this difference was [-0.10-0.02]. These results were the same for both series. This means that at most, the difference in visual acuity measured between the two tests would be 0.10 logMAR.
The results of the two concordance and superiority tests show the same trend: the eMOVA test statistically significantly overestimates visual acuity. This difference, however, is not clinically relevant as it would be at most 0,06 logMAR.
Analysis of the Bland and Altman graphs shows that measurement concordance decreases for low visual acuity. The concordance is good for the visual acuities between 0 and − 0.6 logMAR and becomes lower for the visual acuities lower than − 0.6 logMAR (Fig. 2). This trend is also found for the second series (Fig. 3).
Secondary judgment criteria
The results are shown in Table 3.
Eighty-four percent (84%) of all children had an understanding score of 5. Eighty-one percent (81%) of all children had an attention score of 5. Eighty-four percent (84%) of all children had a distance score greater than or equal to 4. There was no statistically significant difference between the two tests.
The average difference in duration of assessment of visual acuity in both eyes between the two tests is 21 seconds. The minimum duration is 11 seconds for the Rossano-Weiss test and 23 seconds for the eMOVA test. The maximum duration is 340 seconds for the Rossano-Weiss test and 174 seconds for the eMOVA test. These differences are statistically significant (p < 0.001).
Overall, for both tests, we notice that the child's discomfort increases during the test. The average FLACC scores obtained before / during / after the test were all three lower with the eMOVA test. The eMOVA test is less painful for the child than the Rossano-Weiss test and this difference is statistically significant (p = 0.01).
Table 3
Comparison of secondary judgment criteria.
| Rossano-Weiss | eMOVA | p |
Understanding | 4.8 | 4.8 | 0.11 |
Attention | 4.7 | 4.8 | 0.26 |
Respect of distance | 4.4 | 4.3 | 0.72 |
Duration | 43 | 64 | < 0.001 |
Total FLACC | 0.3 | 0.1 | 0.01 |
Test parameters
The reproducibility of the eMOVA test was calculated on a sample of 30 eyes that performed the eMOVA test twice in a row under identical conditions. The reproducibility of the eMOVA test was very good with an intra class correlation coefficient of 0.93 (95% CI [0.87–0.97]). The sensitivity of the eMOVA test was 56% and its specificity 88%. The negative predictive value of the test was 90% and its positive predictive value 53%. The accuracy of the eMOVA test was 82%.
Choice of preferred test
The eMOVA test was mostly chosen by parents and children. Eighty-seven percent (87%) of the children and 80% of the parents chose the eMOVA test.
Sequence effect
No sequence effect was highlighted. There was no statistically significant difference depending on the order of completion of the two tests which limits the bias of our study (p = 0.61).