Screening for Visual Impairment in Children Ages 1–5 Years

Update for the USPSTF

Release Date: January 2011

By Roger Chou, MD; Tracy Dana, MLS; and Christina Bougatsos, BS

The information in this report is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This report is intended as a reference and not as a substitute for clinical judgment.

This report may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.

This report was first published in Pediatrics on January 31, 2011 (Pediatrics 2011;127:e442-e479).

Context: Screening could identify preschool-aged children with vision problems at a critical period of visual development and lead to treatments that could improve vision.

Objective: To determine the effectiveness of screening preschool-aged children for impaired visual acuity on health outcomes.

Methods: We searched Medline® from 1950 to January 2011 and the Cochrane Library through the third quarter of 2009, reviewed reference lists, and consulted experts. We selected randomized trials and controlled observational studies on preschool vision screening and treatments, and studies of diagnostic accuracy of screening tests. One investigator abstracted relevant data, and a second investigator checked data abstraction and quality assessments.

Results: Direct evidence on the effectiveness of preschool vision screening for improving visual acuity or other clinical outcomes remains limited and does not adequately address whether screening is more effective than no screening. Regarding indirect evidence, a number of screening tests have utility for identification of preschool-aged children with vision problems. Diagnostic accuracy did not clearly differ for children stratified according to age, although testability rates were generally lower in children 1 to 3 years of age. Treatments for amblyopia or unilateral refractive error were associated with mild improvements in visual acuity compared with no treatment. No study has evaluated school performance or other functional outcomes.

Conclusions: Although treatments for amblyopia or unilateral refractive error can improve vision in preschool-aged children and screening tests have utility for identifying vision problems, additional studies are needed to better understand the effects of screening compared with no screening.

Visual impairment in young children can reduce quality of life¹ and may affect function and school performance. In the United States, 1% to 5% of preschool-aged children are estimated to have vision impairment that is most commonly related to amblyopia, strabismus, and refractive errors.^2-5 Vision impairment associated with amblyopia is not immediately correctable with refractive lenses, is unlikely to resolve spontaneously,⁶ and can become irreversible.^7,8 Strabismus is the most common amblyogenic risk factor (Table 1). Strabismus can also inhibit development of normal binocular vision in the absence of amblyopia and result in psychosocial consequences.⁹ Preschool vision screening (Table 2), which typically includes a measurement of visual acuity (Table 3), could help identify children who might benefit from early interventions.

In 2004, the US Preventive Services Task Force (USPSTF) recommended screening to detect amblyopia, strabismus, and defects in visual acuity in children younger than 5 years of age ("B recommendation").¹⁴ In 2009, the USPSTF commissioned a new evidence review to update its recommendations. The purpose of this report is to systematically evaluate the current evidence on preschool vision screening.

Using the methods developed by the USPSTF, we developed an analytic framework and key questions (Figure 1) to guide our literature search and review.¹⁵ We searched Ovid Medline from 1950 to July 2009 and the Cochrane Central Register of Controlled Trials and the Cochrane Database of Systematic Reviews through the third quarter of 2009 (Appendix 1). We supplemented electronic searches with reviews of reference lists and by consulting experts.

Figure 2 shows the flow of studies from initial identification of titles and abstracts to final inclusion or exclusion. We selected studies that pertained to screening, diagnosis, and treatment of visual impairment in children 1 to 5 years of age (for details, go to Appendixes 2 and 3). Two reviewers evaluated each study to determine eligibility for inclusion. This review was limited to the published, English-language studies available.

Data from full-text articles were abstracted by 1 investigator and verified by a second investigator. We converted visual acuity from Snellen to logarithmic minimal angle of resolution (logMAR) measurements by using published conversion charts.¹³ Two authors independently rated the internal validity of each study as "good," "fair," or "poor" on the basis of criteria developed by the USPSTF (Appendix 4).^15,16 Discrepancies were resolved by discussion and consensus. For diagnostic accuracy studies, we used the diagti procedure in Stata 10 (Stata Corp, College Station, TX) to calculate sensitivities, specificities, and likelihood ratios. When the reference standard was applied in a random sample of negative screens, we corrected for verification bias by using the method of Begg and Greenes.¹⁷ We classified likelihood ratios as shown in Table 4.¹⁸

We evaluated applicability to populations likely to be encountered in primary care screening settings on the basis of recruitment from primary care settings, the prevalence of visual conditions, and the severity of visual impairment.

We assessed the overall strength of the body of evidence for each key question (good, fair, or poor) by using methods developed by the USPSTF on the basis of the number, quality, and size of studies, consistency of results, and directness of evidence.¹⁵ We did not pool studies of diagnostic test accuracy because of differences in populations, screening cutoffs applied, and target conditions evaluated, as well as between-study heterogeneity in results. There were too few trials of treatments to perform meta-analysis.

This research was funded by the Agency for Healthcare Research and Quality (AHRQ) under a contract to support the work of the USPSTF. AHRQ staff and liaisons from the USPSTF helped develop and refine the key questions and analytic framework and reviewed draft reports. We also distributed an earlier draft of the report for review by external experts who were not affiliated with the USPSTF.

Key Question 1: Is Vision Screening in Children Aged 1 to 5 Years Associated With Improved Health Outcomes?

No randomized trial evaluated preschool vision screening compared with no screening. One large (n = 3490), fair-quality randomized trial nested within a population-based cohort study (the Avon Longitudinal Study of Parents and Children [ALSPAC]) revealed that intensive, periodic orthoptist screening (a clinical examination, age-specific visual acuity testing, and cover-uncover test) from 8 through 37 months of age reduced prevalence of amblyopia at 7.5 years of age by ~1% compared with 1-time screening at 37 months of age, but the difference was only statistically significant for 1 of 2 prestated definitions (Table 5) for amblyopia (amblyopia A, 1.45% vs 2.66%, relative risk [RR]: 0.55 [95% confidence interval (CI): 0.29-1.04]; amblyopia B, 0.63% vs 1.81%, RR: 0.35 [95% CI: 0.15-0.86]).^19,20 Visual acuity at 7.5 years in the amblyopic eye in patched children was better in the intensive-screening group than in the 1-time-screening group by an average of ~1 Snellen line (mean logMAR: 0.15 [95% CI: 0.08-0.22] vs 0.26 [95% CI: 0.17-0.35]; P<0.001). The major methodologic shortcoming of this trial was high loss to follow-up (close to 50%) (Appendix 5).

A large (n = 6081), fair-quality (high-loss-to-follow-up) prospective cohort study from the ALSPAC revealed that 1-time orthoptist screening at 37 months of age was associated with no significant difference in amblyopia risk at 7.5 years compared with school-entry screening when using any of 3 prestated amblyopia definitions (Table 5).²¹ Three poor-quality cohort studies revealed that preschool screening was associated with improved school-aged vision outcomes compared with no screening.^22-24 Besides the use of a retrospective design, methodologic shortcomings in these studies include failure to adjust for potential confounders and varying duration of follow-up. No study evaluated school performance or other functional outcomes.

Key Question 1a: Does Effectiveness of Vision Screening in Children Aged 1 to 5 Years Vary in Different Age Groups?

No randomized trial directly evaluated effectiveness of screening at different age groups in preschool-aged children. The ALSPAC randomized trial initiated screening at different ages (8 vs 37 months), but it is not possible to determine if differences in outcomes should be attributed to the age at which screening was started or the enhanced frequency of screening in the younger group.^19,20 One poor-quality retrospective cohort study of Alaskan children found no significant difference in risk of at least mild vision impairment (visual acuity worse than 20/40) between screening at the age of 2 to 4 years and screening before 2 years of age after 2 to 10 years' follow-up, but estimates were imprecise (RR: 3.10 [95% CI: 0.72-13]).²⁵ In addition, the authors only reported outcomes for 94 children from >10,000 screened and did not adjust for potential confounders. One other retrospective cohort study revealed that the rate of false-positives was approximately twice as high (25% vs 13%) for children screened at 1.5 years compared with those screened at 3.5 years.²⁶

Key Question 2: What Is the Accuracy and Reliability of Risk-Factor Assessment for Identifying Children Aged 1 to 5 Years at Increased Risk for Vision Impairment?

No study evaluated the accuracy or reliability of demographic or clinical features to identify children at higher risk for vision impairment or amblyogenic risk factors before screening, and no study evaluated the yield or outcomes of targeted versus universal preschool vision screening.

Key Question 3: What Is the Accuracy of Screening Tests for Vision Impairment in Children Aged 1 to 5 Years?

Thirty-one studies evaluated the diagnostic accuracy of various preschool vision screening tests (Tables 6 and 7).^27-58 Cycloplegic refraction was included in the reference-standard examination in all but 5 studies.^{28,29,38,40,50} Four studies were rated poor quality,^35,38,45,50 and the other 23 were rated fair quality; the degree to which studies met quality criteria was variable (Appendix 6). The most frequent shortcomings were exclusion of noncompliant children or those with uninterpretable screening tests, failure to describe random or consecutive enrollment of subjects, high or unclear rate of screening failures, and failure to enroll a representative spectrum of subjects.

Nineteen studies evaluated children recruited from pediatric ophthalmology clinics.^{30,31,34-36,38-42,44,45,48,49,52,53,56-58} In these studies, the median prevalence of amblyogenic risk factors was 48% (range: 6%-81%).^{30,31,34,38,39,41,42,44,45,48,49,52,53,57} In 8 studies of children recruited from primary care, community, or school settings, the median prevalence of amblyogenic risk factors was 12% (range: 2%-20%) in 5 studies,^{27,29,37,43,51} and the prevalence of amblyopia was 2% in 3 studies.^28,32,50 The large (n = 2588) Vision in Preschoolers (VIP) Study preferentially enrolled children from Head Start with visual conditions (prevalence of amblyopia: 3%; prevalence of any target visual condition: 29%).^55,60

Visual Acuity Screening

In the VIP Study, crowded Lea symbols visual acuity testing was associated with a positive likelihood ratio (PLR) of 6.1 (95% CI: 4.8-7.6) and negative likelihood ratio (NLR) of 0.43 (95% CI: 0.38-0.50).⁵⁵ A smaller (n = 149) study of children who were attending a pediatric ophthalmology clinic reported moderate-to-strong PLRs (5.7-12) and NLRs (0.05-0.23) depending on the screening cutoff used.³¹ Two studies of Native American children revealed that Lea-symbols testing very weakly increased the likelihood of significant refractive error or astigmatism in high-prevalence settings (PLR: 1.6 and 1.9).^46,47

In the VIP Study, the crowded HOTV test (a test that involves identification of the letters H, O, T, and V) was associated with similar accuracy compared with crowded Lea symbols (PLR: 4.9 [95% CI: 3.9-6.1]; NLR: 0.52 [95% CI: 0.46-0.58]).⁶⁰

Stereoacuity Screening

In 3 fair-quality studies of the random dot E test, the median PLR was 4.2 (range: 3.6-11.4) and the median NLR was 0.65 (range: 0.15-0.81).^32,40,55 The VIP Study had similar results for the random dot E and Stereo Smile II tests (PLR: 4.2 [95% CI: 3.3-5.3] and 4.9 [95% CI: 3.9-6.1]; NLR: 0.65 [95% CI: 0.59-0.71] and 0.62 [95% CI: 0.56-0.67], respectively).⁵⁵

Cover-Uncover Test

In the VIP Study, the cover-uncover test was associated with a PLR of 7.9 (95% CI: 4.6-14) and an NLR of 0.86 (95% CI: 0.82-0.90).⁵⁵

Autorefractors

In 2 studies, the Retinomax autorefractor was associated with a median PLR of 3.4 (range: 1.9-6.1) and median NLR of 0.38 (range: 0.35-0.41).^28,55 From 2 studies of Native American children with astigmatism⁴⁷ or a high prevalence of refractive error,⁴⁶ stronger likelihood ratios (PLR: 6.7 and 18; NLR: 0.11 and 0.08) were reported.

In 3 fair-quality studies, the Sure-Sight autorefractor was associated with a median PLR of 1.8 (range: 1.6-2.2) and median NLR of 0.24 (range: 0.09-0.29) on the basis of the manufacturer's referral criteria.^41,52,55 In the VIP Study, modification of referral criteria to attain a specificity of 0.90 or 0.94 increased the PLR,⁵⁵ but in another study, application of the VIP criteria had little effect on diagnostic accuracy compared with using the manufacturer's criteria.⁵²

In 6 studies of the PlusOptix (previously the Power Refractor), the median PLR was 5.4 (range: 3.0-230) and the median NLR was 0.17 (range: 0.04-0.56).^{27,36-38,45,55} Excluding the poor-quality study³⁸ did not reduce the variability in estimates. One fair-quality study was an outlier, with a PLR of 230 (95% CI: 14-3680).³⁶ Specificity was 100% (252 of 252) in this study, but children with negative screen results did not undergo cycloplegic refraction unless they failed an orthoptist examination. The authors of 1 study reported an improved PLR (from 3.0 to 8.4) when the manufacturer's referral criteria were modified to enhance specificity.⁴⁵

The VIP study revealed slightly stronger likelihood ratios for the Retinomax and SureSight autorefractors compared with the Power Refractor when SureSight screening cutoffs were set to achieve a specificity of 0.90 or 0.94.^55,60

Photoscreeners

In 8 studies of the Medical Technology and Innovations (MTI) photoscreener, the median PLR was 6.2 (range: 2.4-8.7) and the median NLR was 0.26 (range: 0.06-0.67).^{30,35,47,51,52,55-57} One study of Native American children revealed that the MTI photoscreener was associated with a PLR of 2.3 (95% CI: 1.8-2.9) and an NLR of 0.48 (95% CI: 0.38-0.60) for identification of astigmatism (prevalence: 48%).⁴⁷

From 2 studies of the iScreen photoscreener a median PLR of 7.3 (range: 6.2-8.6) and median NLR of 0.25 (range: 0.09-0.67) were reported.^44,55 Two studies of the Visiscreen 100 photoscreener resulted in a median PLR of 7.0 (range: 3.5-14) and median NLR of 0.14 (range: 0.12-0.16).^34,49 Other studies evaluated photoscreeners that are not (or were never) commercially available in the United States.^{35,39,42,43,48} The VIP Study resulted in identical diagnostic accuracy for the MTI and iScreen photoscreeners.⁵⁵

Combinations of Screening Tests

In 5 studies that evaluated combinations of screening tests, the median PLR was 14 (range: 4.8-17) and the median NLR was 0.28 (range: 0.03-0.91).^{29,33,43,50,53} All of the studies included tests of visual acuity, stereoacuity, and ocular alignment, although the specific tests varied.

The VIP study found that addition of an ocular alignment test (cover-uncover test, the Stereo Smile II, or the MTI photoscreener) to a test of visual acuity or refractive error (crowded Lea symbols or HOTV tests or the Retinomax or SureSight autorefractors) increased sensitivity for detection of strabismus by 6% to 31%.⁶¹

Direct Comparisons of Different Types of Screening Tests

The VIP Study found that the random dot E stereoacuity test, the Stereo Smile II, the iScreen photoscreener, and the MTI photoscreener had lower sensitivity compared with crowded Lea symbols or HOTV visual acuity tests and the Retinomax, SureSight, or Power Refractor (PlusOptix), but differences in likelihood-ratio estimates were generally small.⁵⁵ The cover-uncover test was associated with markedly lower sensitivity but higher specificity than the other tests, which resulted in a stronger PLR and weaker NLR.

Key Question 3a: In Children Aged 1 to 5 Years, Does Accuracy of Screening Tests for Vision Impairment Vary in Different Age Groups?

Four studies found no clear differences in the diagnostic accuracy of various screening tests in preschool-aged children stratified according to age (Appendix 7).^33,41,44,56 Testability rates generally exceeded 80% in 3-year-olds, and there were small increases through 5 years of age.^62,65 In the VIP Study, random dot E testability was 86% in 3-year-olds and 93% in 5-year-olds,⁶⁵ and HOTV and Lea-symbols testability was >95% at all ages between 3 and 5 years.⁶⁶ Overall testability was nearly 100% with the MTI photoscreener and various autorefractors.⁵⁵ Most (93%) 3-year-olds in the VIP Study were 42 to 47 months of age, so the applicability of results to younger 3-year-olds is uncertain. Four studies found substantially lower testability (range: 33%-56%) with the random dot E stereotest, Lea symbols, and the SureSight autorefractor in children 1 to <3 years of age compared with those who were older.^41,53,67,68 On the other hand, 1 large study of statewide screening with the MTI photoscreener found that testability was 93% in 1-year-old children.⁶⁹

Key Question 4: What Are the Harms of Vision Screening in Children Aged 1 to 5 Years?

Only 1 controlled study evaluated potential psychosocial effects of screening. In the ALSPAC population-based cohort, children offered screening at 37 months of age were reported to have a 50% decreased odds of being bullied at the age of 7.5 years compared with those who were not offered screening.⁷⁰ Benefits were observed among children who received patching treatment (adjusted odds ratio [OR]: 0.39 [95% CI: 0.16-0.92]) but not among those treated with eyeglasses.

In populations in which the prevalence of visual conditions was <10%, 6^{28,29,32,40,50,53} of 7⁴³ studies that applied the reference standard in all screened children (or a random subset) resulted in false-positive rates of >70% (Appendix 8). One large (n = 102,508) study of a statewide preschool photoscreening program found that 20% of children with positive screen results who did not meet criteria for amblyogenic risk factors were prescribed glasses.⁷¹ In approximately one-quarter of the cases, the refractive error was clinically insignificant (anisometropia ≤0.75 diopter [D], hyperopia ≤2.00 D, myopia ≤0.75 D, and astigmatism ≤0.75 D). The remainder had higher-magnitude refractive errors but did not meet standard criteria for amblyogenic risk factors. No study evaluated the effects of unnecessary corrective lenses or treatment for amblyopia on long-term vision or functional outcomes.

Key Question 5: What Is the Effectiveness of Treatment for Vision Impairment in Children Aged 1 to 5 Years?

In children with unilateral refractive errors, 1 good-quality trial (n = 177) found patching plus eyeglasses and eyeglasses alone each more effective than no treatment by ~1 line on the Snellen eye chart after 1 year (mean difference versus no treatment: 0.11 logMAR [95% CI: 0.05-0.17] and 0.08 logMAR [95% CI: 0.02-0.15], respectively) (Table 8 and Appendix 9). Children were enrolled on the basis of abnormal results of 2 Snellen visual acuity tests but did not necessarily have amblyopia. The average improvement from baseline in logMAR visual acuity was ~0.17 for eyeglasses plus patching, 0.13 for eyeglasses alone, and 0.06 for no treatment, from an average baseline logMAR of 0.36.⁷² In children with moderate (0.48 logMAR or worse) baseline refractive error, patching plus eyeglasses was associated with a larger improvement compared with no treatment (0.27 logMAR [95% CI: 0.14-0.39]).

Two trials evaluated patching versus no patching in children with amblyopia after pretreatment with eyeglasses.^73,74 One good-quality trial (n = 180) by the Pediatric Eye Disease Investigator Group (PEDIG) found that 2 hours daily of patching was associated with improved visual acuity in the amblyopic eye compared with no patching after 5 weeks (mean logMAR: 0.44 [equivalent Snellen 20/50] vs 0.51 [20/63] with no patching; adjusted mean difference: 0.07 [95% CI: 0.02-0.12]).⁷⁴ Forty-five percent of the children in the patching group experienced an improvement of ≥2 lines of visual acuity compared with 23% in the no-treatment group (P =0.003). A smaller, fair-quality trial (n = 60) revealed a trend toward better visual acuity among children (mean baseline log-MAR: 0.64) who were allocated to receive 3 or 6 hours of patching compared with no treatment after 12 weeks (mean change in logMAR: 0.29, 0.34, and 0.24, respectively; P = 0.11 for either treatment versus no treatment).⁷³

All 3 trials evaluated older (4- to 5-year-old) preschool-aged children. No trial evaluated the effects of treatment on school performance or other measures of function.

Five fair- or good-quality trials found no differences in visual acuity improvement in the amblyopic eye between shorter and longer daily patching regimens (2 trials),^75,76 different atropine regimens (2 trials),^77,79 or between patching and atropine (1 trial).⁷⁸

Evidence on whether age affects treatment outcomes is mixed. Two trials^74,75 found no interaction between age and amblyopia treatment effects among preschoolers aged 3 to 7 years, and 1 trial⁷² found delayed treatment for 1 year associated with similar outcomes compared with immediate treatment in children aged 3 to 5 years. A trial of patching versus atropine revealed no interaction between age and visual acuity outcomes in preschoolers aged 3 to 7 years through 2 years of follow-up,^80,81 but at 10 years of age, an age of <5 years at study entry was associated with significantly increased likelihood of amblyopic eye visual acuity of 20/25 or better (57% vs 38%; P =0.004).⁸² One other trial found that younger preschoolers (3 years old) required fewer hours per day of patching to reach significant improvements in visual acuity compared with older preschool-aged children (4-8 years old).⁷⁶

Key Question 6: What Are the Harms of Treatment for Children Aged 1 to 5 Years at Increased Risk for Vision Impairment or Vision Disorders?

Although 1 short-term (5-week) trial found no increased risk of nonamblyopic eye visual acuity loss associated with patching versus no patching,⁷⁴ another trial found patching to be associated with increased risk of ≥2 lines of visual acuity loss compared with the results of atropine (9% vs 1.4%; P < 0.001);⁷⁸ and 1 trial found atropine plus a plano lens to be associated with increased risk of ≥1 line of visual acuity loss compared with the results of atropine alone (17% vs 4%; P =0.005).⁷⁹ In both trials, nonamblyopic eye visual acuity subsequently returned to baseline in almost all children. Two other trials found no difference in risk of nonamblyopic eye visual acuity loss in direct comparisons of different patching or atropine regimens.^75,77

Evidence on adverse psychosocial effects of amblyopia treatments is limited. One fair-quality follow-up study from a randomized trial found that children were more upset by patching plus eyeglasses compared with eyeglasses alone,⁸³ and 1 good-quality trial found patching to be associated with worse emotional well-being compared with atropine.⁸⁴

Results for all key questions are summarized in Table 9.

As in the previous USPSTF review,⁸⁵ direct evidence on improved visual acuity or other health outcomes that result from preschool vision screening remains limited. The only randomized trial to date compared more intensive to less intensive screening rather than screening versus no screening.²⁰ Although it found that repeated preschool screening reduced the prevalence of subsequent (school-aged) amblyopia by ~1% compared with 1-time screening, the difference was only statistically significant for 1 of 2 definitions of amblyopia used in the trial. One fair-quality prospective cohort study found no significant difference between 1-time screening at 37 months of age compared with school-entry screening on risk of amblyopia at 7.5 years of age²¹ but a 50% reduction in the odds of being bullied,⁷⁰ perhaps related to earlier completion of patching regimens. Retrospective cohort studies found preschool vision screening to be more effective than no screening, but they had important methodologic shortcomings.^22-24

More evidence is now available on the accuracy of various preschool vision-screening tests. There is good evidence that commonly used visual acuity tests, stereoacuity tests, the cover-uncover test, autorefractors, and photoscreeners are useful for screening. In the largest study to directly compare many screening tests (the VIP Study), differences in likelihood-ratio estimates were generally too small to clearly distinguish superior from inferior tests.⁵⁵ In addition to diagnostic accuracy, other factors that may affect the choice of screening tests include testability rates at the age being screened, convenience, costs, and how well different tests perform in combination.^{29,33,43,53,61} Screening tests are associated with a high rate of false-positive results in low-prevalence populations,^{28,29,32,40,50,53} which could result in unnecessary prescription of eyeglasses.⁷¹

There is good evidence that there are effective treatments for visual impairment in preschool-aged children. Although benefits of patching compared with no patching averaged ≤1 line of visual acuity, some trials pretreated all children with eyeglasses, and benefits seemed larger (1-2 lines) for children with more severe baseline vision impairment.^72-74 All of the trials enrolled children aged 3 years or older, so applicability to younger preschool-aged children is uncertain. Factors that may affect interpretation of the magnitude of treatment benefits are that the visual impairment associated with amblyopia can become irreversible, is not correctable with refraction, and potentially affects function over the life span of a child.

Evidence on when to initiate preschool screening remains limited. One randomized trial initiated screening at different ages, but the effects of age could not be separated from the effects of repeated versus 1-time screening.²⁰ Results of other studies indicate a lower rate of false-positive screens in children screened at 3.5 years compared with those screened at 1.5 years²⁶ and no clear association between age at which treatment was started and effectiveness among preschool-aged children aged 3 years and older.^{72,74-76,80-82}

Our evidence review has some potential limitations. First, we excluded non-English-language studies, which could introduce language bias. However, we identified no relevant non-English-language studies in our literature searches. Second, there were too few studies to assess for publication bias. Third, a number of studies evaluated diagnostic accuracy of screening tests or screening programs in community-based settings and specialty eye clinics, which could limit their applicability to primary care settings.

Well-designed studies are needed to better understand the effects of screening compared with no screening, to identify optimal methods for vision screening, to clarify when to begin screening, to define appropriate screening intervals, and to develop effective strategies for linking preschool-aged children with vision impairment to appropriate care while avoiding unnecessary use of eyeglasses and other treatments. In addition, almost all of the trials have focused on effects of preschool vision screening and treatment on visual acuity. Trials that also address function are needed to clarify how preschool vision screening may affect school performance and other aspects of child development.

Direct evidence on the effectiveness of preschool vision screening for improving visual acuity or other clinical outcomes remains limited and does not adequately address the question of whether screening is more effective than no screening. However, good evidence on diagnostic accuracy and treatments suggest that preschool vision screening could lead to increased detection of visual impairment and greater improvement in visual outcomes than if children were never screened.

This document is in the public domain within the United States.

Requests for Single Reprints: Reprints are available from the USPSTF Web site (www.uspreventiveservicestaskforce.org).

Source: Chou R, Dana T, Bougatsos C. Screening for visual impairment in children ages 1-5 years: update for the USPSTF. Pediatrics 2011;127:e442-e479.

Acknowledgments: This work was conducted by the Oregon Evidence-Based Practice Center under contract to the Agency for Healthcare Research and Quality, Rockville, MD (contract HHSA-290-2007-10057-I-EPC3, task order 3).

The authors acknowledge Rongwei Fu, PhD (Oregon Health & Science University), for statistical assistance; pediatric ophthalmologist David Wheeler, MD (Casey Eye Institute); the expert reviewers of the draft report; Agency for Healthcare Research and Quality Medical Officer Iris Mabry-Hernandez, MD, MPH; and USPSTF leads David Grossman, MD, MPH, Thomas G. DeWitt, MD, Virginia Moyer, MD, MPH, and Bernadette Melnyk, PhD, RN, for their contributions to this report.

Disclaimer: Recommendations made by the USPSTF are independent of the U.S. government. They should not be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.

D = diopter.
* Anisometropia is a difference in the refractive power of the two eyes.
Reprinted with permission from Donahue et al.¹⁰ (p 315).

Data sources: American Academy of Pediatrics¹¹ and Prevent Blindness America.¹²

Visual impairment is 20/50 or worse; legal blindness is 20/200 or worse.
Data source: Holladay.¹³

NTU indicates National Taiwan University; TNO, a Dutch stereoacuity test.

^a Based on 90% specificity.
^b Based on 0.80 acuity score cutoff.
^c Values are median (range).
^d Excluded from calculation of median.
^e CIs not calculable.
^f Based on manufacturer's referral criteria.
^g Based on VIP Study 90% specificity criteria.
^h Based on median results from multiple readers (numbers in parentheses are ranges).

ANOVA indicates analysis of variance.

Overall Searches

Database: EBM Reviews--Cochrane Central Register of Controlled Trials

1. amblyopia.mp. [mp = title, original title, abstract, Medical Subject Headings (MeSH), heading words, key word]
2. strabismus.mp. [mp = title, original title, abstract, MeSH, heading words, key word]
3. refractive error.mp. [mp = title, original title, abstract, MeSH, heading words, key word]
4. 1 or 2 or 3
5. 4 and (child$ or pediatri$ or preschool). mp. [mp = title, original title, abstract, MeSH, heading words, key word]
6. limit 5 to yr = "2003-2008"

Database: EBM Reviews--Cochrane Database of Systematic Reviews

1. amblyopia.mp. [mp = title, short title, abstract, full text, key words, caption text]
2. strabismus.mp. [mp = title, short title, abstract, full text, key words, caption text]
3. refractive error.mp. [mp = title, short title, abstract, full text, key words, caption text]
4. 1 or 2 or 3
5. 4 and (child$ or pediatri$ or preschool). mp. [mp = title, short title, abstract, full text, key words, caption text]

Risk Search

Database: Ovid Medline

1. exp Amblyopia/
2. exp Refractive Errors/
3. exp Vision Disorders/
4. or/1-3
5. limit 4 to ("newborn infant [birth to 1 month]" or "infant [1-23 months]" or "preschool-aged child [2-5 years]")
6. exp Risk/ or exp Risk Factors/
7. 5 and 6
8. limit 7 to yr = "1999 -2008"
9. Case Reports/
10. 8 not 9

Screening Search

Database: Ovid Medline

1. vision tests/ or refraction, ocular/ or vision screening/
2. limit 1 to ("newborn infant [birth to 1 month]" or "infant [1-23 months]" or "preschool-aged child [2-5 years]")
3. limit 2 to yr = "1999 -2008"
4. limit 3 to humans
5. limit 4 to English language
6. limit 4 to abstracts
7. 5 or 6
8. Case Reports/
9. 7 not 8
10. English abstract.mp.
11. 9 not 10

Treatment Search

Database: Ovid Medline

1. exp Amblyopia/dt, pc, th [Drug Therapy, Prevention & Control, Therapy]
2. exp Refractive Errors/dt, th, pc [Drug Therapy, Therapy, Prevention & Control]
3. 1 or 2
4. limit 3 to ("newborn infant [birth to 1 month]" or "infant [1-23 months]" or "preschool-aged child [2-5 years]")
5. limit 4 to English language
6. limit 4 to abstracts
7. 5 or 6
8. limit 7 to yr = "1999-2008"

We defined the target population as children 1 to 5 years of age who were evaluated in primary care or community-based settings without known visual impairment or obvious symptoms. We included studies of screening in specialty eye settings but evaluated their applicability to primary care settings. Although the term "visual impairment" is broad, conditions covered in this review are amblyopia, amblyogenic risk factors (Table 1), strabismus, and simple refractive errors.

For screening tests, we included tests of visual acuity, ocular alignment, and stereoacuity; photoscreeners; and autorefractors. Preschool vision screening typically includes measurement of visual acuity (Table 3), ocular alignment, and stereoacuity.⁸⁶ Recommended visual acuity tests vary according to age (Table 2). Potential advantages of more automated screening methods such as photoscreeners and autorefractors over traditional screening are that they may reduce testing time, increase objectivity of screening, or enhance testability rates in younger children. Children who fail a preschool vision-screening test are typically referred for a full ophthalmologic examination to confirm the presence of vision problems. We excluded tests not commonly used in primary care, including cycloplegic refraction and retinoscopy.

For treatments, we focused on correction of refractive errors and use of patching or atropine. Outcomes of interest were visual acuity, risk of amblyopia, vision-related function, school performance, and adverse events related to screening or treatment. We excluded children with severe congenital conditions or developmental delay, retinopathy of prematurity, glaucoma, congenital cataract, and high myopia.

All Key Questions

Ages

• Include children aged 1 to 5 years
• Exclude newborns and children younger than 1 year of age and children aged 6 years and older

Diseases

• Include amblyopia, amblyogenic risk factors, refractive error
• Exclude children with severe congenital conditions or developmental delay, retinopathy of prematurity, glaucoma, congenital cataract, pathologic myopia

Language/Publication Status

• Include full-text (ie, not available only as a conference abstract) journal article published in English

Settings

• Include studies performed in primary care, community-based, and school settings
• Exclude countries with populations not similar to that of the United States

Study Designs

• Exclude systematic reviews

Key Questions 1 (Screening and Outcomes) and 1a (Variation in Age Groups)

Interventions/Diagnostic Tests

• Include studies of screening tests used or available in primary care settings (eg, visual acuity tests, tests of stereopsis, tests for strabismus, photoscreeners, autorefractors)
• Exclude studies of screening tests not used or available in primary care settings (eg, contrast sensitivity testing, fundoscopic examination, visual acuity testing with cyclopegia) or not intended to detect amblyopia, amblyogenic risk factors, or refractive errors (eg, white reflex screening)

Outcomes

• Include visual acuity, long-term amblyopia, school performance, function, quality of life

Study Designs

• Include randomized controlled trials (RCTs) and controlled observational studies

Key Question 2 (Accuracy/Reliability of Risk-Factor Assessment)

Outcomes

• Include studies on accuracy or yield of risk-factor assessment for targeted screening, or clinical outcomes associated with use of targeted versus universal screening

Study Designs

• Include RCTs and controlled observational studies

Key Questions 3 (Accuracy of Screening Tests) and 3a (Variation in Age Groups)

Diagnostic Tests

• Include studies of screening tests used or available in primary care settings (eg, visual acuity tests, tests of stereopsis, tests for strabismus, photoscreeners, autorefractors)
• Exclude studies of screening tests not used or available in primary care settings (eg, contrast sensitivity testing, fundoscopic examination, visual acuity testing with cyclopegia) or not intended to detect amblyopia, amblyogenic risk factors, or refractive errors (eg, white reflex screening)

Outcomes

• Include sensitivity, specificity, positive and negative predictive values, likelihood ratios, diagnostic ORs (or able to calculate such outcomes from data provided)

Study Designs

• Include studies on diagnostic accuracy of a screening question or diagnostic test compared with a credible reference standard (ie, cycloplegic refraction)
• Exclude studies that do not attempt to perform the reference standard in all patients, or a random sample

Key Question 4 (Harms of Screening)

Interventions/Diagnostic Tests

• Include studies of screening tests used or available in primary care settings (eg, visual acuity tests, tests of stereopsis, tests for strabismus, photoscreeners, autorefractors)
• Exclude studies of screening tests not used or available in primary care settings (eg, contrast sensitivity testing, fundoscopic examination, visual acuity testing with cyclopegia) or not intended to detect amblyopia, amblyogenic risk factors, or refractive errors (eg, white reflex screening)

Outcomes

• Include harms, including psychological distress, labeling, anxiety, other psychological effects, false-positive results, adverse effects on nonimpaired eye vision

Study Designs

• Include RCTs and controlled observational studies

Key Question 5 (Effectiveness of Treatment)

Interventions/Treatments

• Include correction of refractive errors (eyeglasses), patching, and atropine

Outcomes

• Include visual acuity, long-term amblyopia, school performance, function, quality of life

Study Designs

• Include RCTs

Key Question 6 (Harms of Treatment)

Interventions/Treatments

• Include correction of refractive errors and penalization of the nonamblyogenic eye (patching and atropine)

Outcomes

• Include harms, including psychological distress, labeling, anxiety, other psychological effects, false-positive results, adverse effects on nonimpaired eye vision

Study Designs

• Include RCTs and controlled observational studies

Diagnostic Accuracy Studies

Criteria

• Screening test relevant, available for primary care, adequately described
• Study uses a credible reference standard, performed regardless of test results
• Reference standard interpreted independently of screening test
• Handles indeterminate results in a reasonable manner
• Spectrum of patients included in study
• Sample size
• Administration of reliable screening test
• Random or consecutive selection of patients¹⁶
• Screening cutoff predetermined¹⁶
• All patients undergo the reference standard¹⁶

Definition of Ratings Based on Criteria Listed Above

Good: Evaluates relevant available screening test; uses a credible reference standard; interprets reference standard independently of screening test; reliability of test assessed; has few or handles indeterminate results in a reasonable manner; includes large number (>100) of broad-spectrum patients with and without disease; study attempts to enroll a random or consecutive sample of patients who meet inclusion criteria¹⁶; screening cutoffs prestated¹⁶

Fair: Evaluates relevant available screening test; uses reasonable although not best standard; interprets reference standard independent of screening test; moderate sample size (50-100 subjects) and a "medium" spectrum of patients (ie, applicable to most screening settings)

Poor: Has important limitation such as uses inappropriate reference standard; screening test improperly administered; biased ascertainment of reference standard; very small sample size of very narrowly selected spectrum of patients

RCTs and Cohort Studies

Criteria

• Initial assembly of comparable groups: RCTs--adequate randomization, including concealment and whether potential confounders were distributed equally among groups; cohort studies--consideration of potential confounders with either restriction or measurement for adjustment in the analysis; consideration of inception cohorts
• Maintenance of comparable groups (includes attrition, cross-overs, adherence, contamination)
• Important differential loss to follow-up or overall high loss to follow-up
• Measurements: equal, reliable, and valid (includes masking of outcome assessment)
• Clear definition of interventions
• Important outcomes considered
• Analysis: adjustment for potential confounders for cohort studies, or intention-to-treat analysis for RCTs; for cluster RCTs, correction for correlation coefficient

Definition of Ratings Based on Criteria Listed Above

Good: Meets all criteria: comparable groups are assembled initially and maintained throughout the study (follow-up at least 80%); reliable and valid measurement instruments are used and applied equally to the groups; interventions are spelled out clearly; important outcomes are considered; and appropriate attention to confounders in analysis

Fair: Studies will be graded "fair" if any or all of the after problems occur, without the important limitations noted in the "poor" category below: generally comparable groups are assembled initially, but some question remains whether some (although not major) differences occurred in follow-up; measurement instruments are acceptable (although not the best) and generally applied equally; some but not all important outcomes are considered; and some but not all potential confounders are accounted for

Poor: Studies will be graded "poor" if any of the following major limitations exists: groups assembled initially are not close to being comparable or maintained throughout the study; unreliable or invalid measurement instruments are used or not applied at all equally among groups (including not masking outcome assessment); and key confounders are given little or no attention

Case-Control Studies

Criteria

• Accurate ascertainment of cases
• Nonbiased selection of cases/controls with exclusion criteria applied equally to both
• Response rate
• Diagnostic testing procedures applied equally to each group
• Measurement of exposure accurate and applied equally to each group
• Appropriate attention to potential confounding variable

Definition of Ratings Based on Criteria Listed Above

Good: Appropriate ascertainment of cases and nonbiased selection of case and control participants; exclusion criteria applied equally to cases and controls; a response rate of ≥80%; diagnostic procedures and measurements accurate and applied equally to cases and controls; and appropriate attention to confounding variables

Fair: Recent, relevant, without major apparent selection or diagnostic workup bias but with a response rate of <80% or attention to some but not all important confounding variables

Poor: Major selection or diagnostic workup biases, response rates of <50%, or inattention to confounding variables

Sources: Harris et al¹⁵ and Leeflang et al.¹⁶

NA indicates not applicable.

a CIs not calculable.

TNO indicates a Dutch stereoacuity test.

^a Raw data not provided; unable to calculate CIs.
^b Calculation based on n = 379, median sensitivity and specificity.
^c Based on reported sensitivity and specificity; do not match values reported in article.

a Converted from Snellen metric measures.

Key questions:

1. Is vision screening in children aged 1-5 y associated with improved health outcomes?
   1a. Does effectiveness of vision screening in children aged 1-5 y vary in different age groups?
2. What is the accuracy and reliability of risk-factor assessment for identifying children aged 1-5 y at increased risk for vision impairment?
3. What is the accuracy of screening tests for vision impairment in children aged 1-5 y?
   3a. In children aged 1-5 y, does accuracy of screening tests for vision impairment vary in different age groups?
4. What are the harms of vision screening for children aged 1-5 y?
5. What is the effectiveness of treatment for vision impairment in children aged 1-5 y?
6. What are the harms of treatment for children aged 1-5 y at increased risk for vision impairment or for vision disorders?

a Cochrane databases include the Cochrane Central Register of Controlled Trials and the Cochrane Database of Systematic Reviews.
b Other sources include reference lists, suggested by peer reviewers, etc.
c Some articles are included for more than 1 key question.