The Norwegian Directorate of Health are updating their guideline on vision screening for children under the age of 18, and have therefore commissioned this systematic overview which aim is to summarise, and critically appraise, evidence from studies that compare the effect of screening with no screening (or screening of different intensity).
We included five studies (one non-randomised controlled trial with a follow up study, and three cohort studies), including a total of 18,497 children, who were aged 6 months to 8 years at follow up. Screening history, screening tests used, and timing of screening varied across studies. Only single studies contributed data to each comparison. No study evaluated the effects of school-screening.
- It is uncertain whether vision screening of children results in more amblyopia cases being identified, as compared to no screening or fewer screenings (4 studies; very low-certainty evidence).
- Vision screening of preschoolers may possibly result in more visual deficits being identified, as compared to no screening (1 study; low- certainty evidence).
- It is uncertain whether vision screening of children results in improved referral, as compared to no screening or fewer screenings (2 studies; very low-certainty evidence).
- It is also uncertain whether vision screening of children results in improved treatment outcomes, as compared to no screening (one study; very low-certainty evidence).
We cannot, based on the very low to low certainty of evidence from the few studies included in this review, draw any firm conclusions on the effects of vision screening in children, or of different screening intensity. Future studies should use a randomised study design, or if that is not feasible, use consecutive birth cohorts to ensure comparability of groups.
There is a lack of consensus concerning the effects, and cost-effectiveness, of vision screening in children, as well as the optimal age, the frequency and the intervals, at which to carry out screening. In Norway, all children are screened at birth, at 6 weeks of age, at 3 months, and at the age of 4 at the mandatory health checks. In contrast to guidance in the other Scandinavian countries, the Norwegian Guideline for Vision Screening in children recommends fewer pre-school screening events, and does not include any recommendations for screening of schoolchildren. Vision screening of children at different ages fulfil different purposes, e.g. preschool screening to detect amblyopia (lazy eye), and related conditions, and screening of schoolchildren to detect refractive errors (blurred vision), and progressive visual deficits. Younger children are at risk of permanent impaired vision if deficits are not identified in time, while older children do not risk permanent visual impairment if not screened. Uncorrected visual deficits in young children may have a negative impact not only on children’s sensorimotor development, but also on their intellectual and social development. This report is a systematic review of the effects of vision screening in children on the detection of disease and treatable deficits, referral to appropriate treatment, and follow up of the children’s vision.
The overall objective of this systematic review was to summarise and critically appraise the existing evidence for the effects of vision screening in children under the age of 18, on the detection of disease and treatable visual deficits, referral to adequate treatment, and follow up of these children.
We conducted a systematic review in accordance with the Cochrane handbook and the handbook of the Division of Health Services at the Norwegian Institute of Public Health. We searched for studies in six electronic databases up to May 2018. Two authors independently screened all titles, and thereafter assessed the full texts of possible eligible studies. One review author extracted data onto a standardised data extraction form, and a second review author checked the accuracy of the data. The same two authors independently assessed the quality of the included studies using the Cochrane EPOC risk of bias tool for non-randomised controlled trials, and ROBINS-1 tool for cohort studies. We assessed the certainty of the included evidence using the GRADE tool (Grading of Recommendations Assessment, Development and Evaluation). Disagreements were solved through discussion between authors.
We included five heterogeneous studies in this systematic review. These studies (one non-randomised controlled trial with a follow up study, and three cohort studies) provided data for five comparisons, i.e. only single studies contributed data for each comparison. All studies evaluated the effect of screening of young children (i.e. before school start), and none evaluated the effects of school-screening. The studies reported three primary outcomes: amblyopia prevalence/visual deficits (5 studies); referrals (2 studies) and treatment outcomes (one study). The certainty of the evidence for all the primary outcomes was overall very low.
Comparison 1: Vision screening at 1-2.5 years of age vs. no screening: outcomes assessed at 8 years of age (one study)
One study reported fewer amblyopia cases among 8-year old children previously screened at 1-2.5 years of age (8/808; 1.0%), as compared to previously unscreened children (20/782; 2.6%; RR (95% CI): 0.39 (0.17 to 0.87); P=0.0098) .
The same study also reported lower prevalence of more severe amblyopia in previously screened children (1/808; 0.1%), as compared to in unscreened children (13/78; 21.7%); RR (95% CI): 0.07 (0.01 to 0.57); P=0.01. The authors did not provide any information on the number of children in the unscreened group who had been diagnosed with amblyopia and treated. This study also reported referrals, number of children diagnosed and treated for amblyopia, but only for the screened group.
Comparison 2: Screening at 4-4.5 years of age vs. no screening: effects assessed at 5.5 years of age
One study  reported that fewer of the children who had been screened 6-12 months previously had visual deficits (78/763 children; 10%), as compared to unscreened children (112/743; 15%; RR (95% CI): 0.68 (0.52 to 0.89); P=0.005. The authors provided no further information on the type of visual deficits, apart from the severity of deficits.
Comparison 3: Screening 6-9 months vs. omitting screening at 6-9 months: effects assessed at 9 months
One study  reported a similar proportion of children that after referral were diagnosed with amblyopia combined with strabismus in the screened group (10/ 6,059; 0.17%) as in the unscreened group (6/5,482 infants; 0.11%); RR (95% CI): 1.51 (0.55 to 4.15), P=0.43.
The same study reported a similar proportion of referrals among infants screened at 6-9 months (58/6,059; 0.96%), as among those with no screening at 6-9 months (48/5,482; 0.88%); RR (95% CI): 1.09 (0.75 to 1.61), P=0.65. However, the confidence interval around the effect estimate, for both amblyopia prevalence and referrals, was wide, with the effect ranging from being in favour of the screening to a non-favourable effect.
Comparison 4: Intensive screening (at 8, 12, 18, 25, and 31 months) vs. no screening (visual surveillance only): effects assessed at 37 months
One study  reported that more toddlers with amblyopia were identified in the intensive screening programme (33/2,029; 1.6 %), as compared to in the unscreened group (8/1,461; 0.5%); RR (95% CI): 2.97 (1.38 to 6.41), P<0.006. Note that the results, which refer to an increased number of amblyopia cases identified through intensive screening, as compared to visual surveillance only (by health visitors and GPs), was considered a desired result.
The same study  reported in total fewer referrals among children in the intensive screening group (147/2,029; 7.2%), as compared to among control group participants (135/1,461; 9.2%); RR (95% CI):0.77 (0.60 to 0.98), P=0.03.
Comparison 5: Intensive screening (at 8, 12, 18, 25, and 31 months) vs. no screening (visual surveillance only): effects assessed at 7.5 years of age (follow-up study)
Fewer amblyopia cases were reported at follow up , among 7.5 year old children in the intensive screening group (22/1088; 2.0%), than in the group of unscreened children (37/826; 4.48%); RR (95% CI): 0.45 (0.27 to 0.76), P=0.003. A problem with this study was large losses to follow up.
The same study  also reported that a similar proportion in both groups received patches (intensive screening: 40/1088; 3.7% vs. unscreened: 40/826¸4.8%); RR (95% CI): 0.76 (0.49 to 1.17), P=0.21, and fewer cases with residual amblyopia in the intensive group (3/40), as compared to in the unscreened group (10/40); RR (95% CI): 0.30 (0.09 to 1.01), P=0.05. In addition, the mean visual acuities (in the worse seeing eye) was better for children in the intensive group than for similar children in the control group (mean acuity 0.15 (95% CI; 0.085 to 0.215) vs. 0.26 (0.173 to 0.347) LogMAR units; P < 0.001).
The certainty of evidence from the five included studies and for the primary outcomes (amblyopia prevalence, referrals and treatment outcomes) was very low to low, which means that we have low confidence in the results.
We included only five heterogeneous non-randomised studies in this systematic review (one NRCT with a follow up study, and three controlled cohort studies), which all evaluated the effect of vision screening of younger children, and none the effect of school-screening.
The included studies were heterogeneous in terms of populations, timing of the screening, screening history of the children, screening tests used and in the profession of the screeners. This, and the fact that only single studies provided data for each comparison prevented meta-analysis.
The certainty of the included evidence was very low to low, and it is therefore not possible to draw any firm conclusion on the effects of screening as compared to no screening, or to different screening intensity.
The results are based on very low to low-certainty evidence from five heterogeneous non-randomised studies, and we cannot therefore draw any firm conclusions on whether or not vision screening in children lead to improved eye health (i.e. lower prevalence of amblyopia and other vision deficits), as compared to no screening (or fewer screenings). Future studies should use a randomised study design, or if that is not feasible, use consecutive birth cohorts to ensure the comparability of groups.