Trager and associates recently reported interesting findings from the Strabismus, Amblyopia and Refractive Error Study (STARS) on the testability of vision and refraction in a Singaporean sample of Chinese preschool children, aged 6 to 72 months. The authors correctly indicated that there was a paucity of studies in the literature on the testability of different vision screening modalities in preschool children internationally, and that most publications were based on relatively small sample sizes. This makes it difficult to evaluate the sensitivity, the specificity, and the practical values of some vision testing methods in preschool populations.

The Sydney Paediatric Eye Disease Study (SPEDS), currently ongoing in Australia, has recruited and examined 2473 preschool children aged between 6 and 72 months, from various ethnicities. Visual acuity (VA) testing was performed using either logarithm of the minimal angle of resolution (logMAR) or electronic visual acuity (EVA). Refraction was performed with both the hand-held Retinomax K-Plus 2 autorefractor and the table-mounted Canon RK-F1 autorefractor when possible, or by streak retinoscopy if both autorefraction measurements failed. Randot stereoacuity was also attempted on all participants aged over 30 months. Similarly, IOL Master measurements were attempted on all children, also mainly aged over 30 months, using comparable methodology to that adopted by STARS. Testability was defined as successful measurements in both eyes for VA, refraction, stereopsis, and at least 2 axial length measurements.

Our preliminary findings on VA testing with logMAR and EVA were recently presented (Leone J, et al. IOVS 2009;50:ARVO E-abstract 1583). Compared to the Singapore cohort, we had lower testability using logMAR, including the oldest age group. However, our testability using EVA was higher than using logMAR, and was comparable to VA testability in STARS, reported by Trager and associates. Examiner variability is 1 potential reason for the VA testability difference between studies. Another possible explanation that should be considered is the relatively high testability using EVA in young children, attributable to its single-surround letter design. After adequate cycloplegia (pupil diameter at least 6 mm), the testability using the Canon autorefractor showed similar findings between the 2 cohorts, including the youngest age group. In contrast to STARS findings, however, we found that girls were not significantly more testable using table-mounted autorefraction than boys, after adjusting for age (70.6% vs 64.1%, P = .062). Interestingly, our cohort was more testable using Randot stereoacuity for all age groups than the Singapore cohort. With Randot, we found no significant gender differences (age-adjusted P = .87), nor did we find any significant difference in testability between children of Caucasian and East Asian ethnicity ( P = .35). Singaporean preschool children outperformed the Sydney cohort in the testability using IOL Master. Using IOL Master, we found no significant gender differences after adjusting for age ( P = .85), similar to the Singapore group, nor did we find a difference between children of Caucasian and East Asian ethnicity ( P = .39). Finally, as expected and consistent with the authors’ findings, we found in our preliminary data that the testability of our preschool population for VA, refraction, stereoacuity, and IOL Master improved with increasing age ( P trend < .0001).