This was a prospective trial conducted in two phases as part of a school-based vision screening program from March–May 2015. The study population included children aged 4 to 14 years who were enrolled in full-time education at nine Hamilton-Wentworth Catholic District School Board (HWCDSB) elementary schools. Following parental consent, each child received both a full eye examination by the study optometrist and vision screening by one of the eight volunteer examiners (See Fig. 1). All children received a full eye examination by the study optometrist which included a dilated fundus examination and refraction. Full eye exams were provided by the optometrist, rather than screening, as an added benefit to study participation to increase consent rate. Research ethics approval was obtained from both the Hamilton Integrated Research Ethics Board and the Research Advisory Committee of the HWCDSB. All research conducted adhered to the tenets of the Declaration of Helsinki.
Phase I: Training the vision screeners
Eight undergraduate bachelor’s degree students (four health sciences students, two nursing students, and one each from arts and life sciences) underwent 40 h of training to correctly perform the following tests: distance visual acuity (VA) using Snellen crowded letters or LEA symbols (depending on child’s age); near VA using Rosenbaum chart or LEA symbols; Ishihara color vision; and Randot stereoacuity. All training was conducted by the study optometrist. Volunteer examiners were educated, trained and assessed on the principles and practices of performing vision screening tests, using videos, written instructions, practical exercises and supervised eye examinations, all of which took place at the paediatric eye clinic at McMaster Children’s Hospital in Hamilton, Ontario, Canada.
Phase II: In-school screening
Research was conducted at nine local Catholic schools in Hamilton (Ontario, Canada) over the course of 8 weeks, including children between 4 and 14 years of age. Schools were randomly selected to be approached for inclusion, and principals had to agree to participate prior to on-site visits. Each child received a full eye exam by the optometrist and vision screening performed by volunteer screeners, in randomly selected sequence. If the optometrist was randomly designated to examine the child first, they performed all tests exclusive of dilated fundus examination and refraction, then completed that portion of the exam after the child had been screened by the volunteers. The M&S Smart System (M&S Technologies, Niles, IL), a computer-based visual acuity testing system where the Snellen chart is shown on a computer screen, was used for measuring VA. Visual Acuity was recorded as the lowest line on which the child correctly identified half or more of the optotypes while the fellow eye was occluded with an adhesive Ortopad orthoptic eye patch (Ortopad USA, Tuscon, AZ). Each child’s distance visual acuity was measured at a distance of 20 ft with Snellen crowded letters, or LEA symbols if the child could not read. Monocular near visual acuity was measured using the Rosenbaum chart at 36 cm or LEA Symbol chart at 40 cm. It was ensured that for any given child, the same test was used by both the optometrist and volunteer for distance visual acuity (Snellen or LEA) and near visual acuity (Rosenbaum or LEA) measurement. Ishihara plates were used for monocular color vision testing and the Randot stereotest was administered for near stereoacuity. Colour vision was scored separately for each eye as the number of plates the child identified correctly. Assessment of colour vision was included in the tests conducted in this study as upwards of 8% of the global male population experiences colour deficiency. Defects in colour vision can often go undetected and may affect a child’s learning . Assessing children’s stereopsis was included in the study as an indirect way of detecting children with manifest strabismus. In cases of manifest strabismus, even if the visual acuity is good in both eyes, the stereopsis will be reduced. Therefore, by testing stereopsis, the volunteer examiners would be able to detect children with suspected strabismus. Stereoacuity tests are part of the screen and refer processes in some vision screening programs in both the US and Canada [20, 21]. If the child wore glasses, all vision screening assessments were performed with the child wearing his or her glasses. None of the children in this study wore contact lenses.
Data collection and data standardization
Study data were entered into a password protected, encrypted computer database. Data collected included gender, age, school, near visual acuity, distance visual acuity, stereoacuity and color vision.
Sample size calculation and data analysis
A previously published rate of impaired vision in Hamilton area elementary schools was used to calculate sample size. Assuming that approximately 20% of children enrolled will have vision impairment and an alpha error of 0.05, we needed 353 participants to achieve a confidence interval of + 5% around the expected sensitivity and specificity of 90%.
We decided on an a priori moderate to high level of agreement as acceptable for all phases of the study (ICC of 0.61 or greater) . Intraclass correlation coefficient (ICC) was calculated using two-ways mixed effects models for overall reliability of the volunteer examiners to conduct vision tests compared to the optometrist. ICCs as a measure of reliability with the corresponding 95% confidence intervals (CI) are reported. Sensitivity, specificity, and likelihood ratios with corresponding 95% CIs were reported to measure accuracy of volunteer examiners’ vision screening compared to the study optometrist. SPSS (www.ibm.com) was used for data analysis.