Methodological quality of the review: Medium confidence
Author: Carlton J, Karnon J, Czoski-Murray C, Smith KJ, Marr J.
Region: United Kingdom (UK), United States of America (USA), The Netherlands, Brazil, Canada, Belgium, Germany, Israel
Sector: Amblyipia, strabismus
Sub-sector: Effectiveness, cost-effectiveness, screening programmes
Equity focus: None specified
Review type: Effectiveness review
Quantitative synthesis method: Narrative analysis
Qualitative synthesis methods: Not applicable
Amblyopia and strabismus are both common conditions in childhood. A Health Technology Assessment report published in the UK in 1997 concluded that the evidence for the value of screening for such conditions did not support any expansion of the current screening programme; it recommended that the National Screening Committee should consider halting the existing programme. The authors specifically highlighted the lack of evidence on the long-term impact of amblyopia, the extent of disability that amblyopia and strabismus have and their impact of quality of life.
This study aimed to re-examine the literature and the use of this to inform a decision-analytic model to determine the cost-effectiveness of screening for amblyopia and strabismus.
The specific objectives of the study were:
A total of 19 papers reported data on screening that evaluated specific tests that could be used to screen children for amblyopia, strabismus and/or amblyogenic factors: 12 papers examined screening programmes; six examined who should undertake the vision screening; and seven papers reported on the impact of screening programmes on treatment outcomes.
The amblyopia screening model was analysed in detail to estimate the cost and effects of six alternative screening options comprising screening at different ages (three, four and five years) and using alternative sets of tests (visual acuity testing and cover tests, with and without auto refraction). The reference case results showed that screening programmes that included auto refraction dominated screening programmes without auto fraction. Analyses based on the cost per case of amblyopia prevented showed that screening at either three or four years prevented additional cases at a low absolute cost (£3000-6000). However, when these results were extrapolated to estimate the cost per quality-adjusted life-year (QALY) gained, the reference case analysis found that no form of screening is likely to be cost-effective at currently accepted values of a QALY.
The wide-ranging sensitivity analyses found that the results were robust to most parameter changes. The only parameter that radically affected the results was the utility effect of loss of vision in one eye. No direct evidence of a utility effect was identified and the reference case assumed (a reduction in utility of 2%), the incremental cost per QALY gained becomes extremely attractive for screening at both three and at four years. The expected value of perfect information was shown to be large when the unilateral vision loss utility parameter was allowed to vary, but not when it was kept constant at zero.
Separate systematic reviews were undertaken to determine the prevalence and natural history, the screening methods, the effectiveness of treatment options and health-related quality of life issues relating to amblyopia and strabismus. It should be noted that the full expected value of perfect information (EVPI) analysis was undertaken for amblyopia and strabismus in the UK. Inclusions criteria for all systematic reviews consisted of primary research, systematic review or high-quality review which reported data relevant to the systematic review, nevertheless, it should be noted that that for the prevalence and incidence systematic review included only UK-based data.
The authors conducted a search on the following databases: Cochrane Database of Systematic Reviews, Central, Embase, Medline, Biosis, Cinahl, NHS EDD, OHE HEED, Science Citation Index, NHS DARE and NHS HTA. In attempt to identify grey literature, the authors searched appropriate databases such as Health Management Information Consortium, current research registers and relevant databases. The reference of included studies and review articles were also checked. No date or language restrictions were applied to the searches.
Data were extracted by one reviewer and checked by a second review using specifically designed data extraction tables. The QUOROM statement checklist for reporting systematic reviews was used and presented as an appendix.
The authors noted that the papers describing the impact of screening programmes differed widely in the content of screening programme itself, the population group examined and the personnel administering the screening.
Only studies from high-income countries were included in the review, and as such, findings reported in the review may only be applicable to high-income settings only.
For the prevalence studies systematic review, all non-UK-based data was excluded.
Overall, there is medium confidence in the conclusions about the effects of this study. The search for literature was thorough to ensure the inclusion of all relevant studies avoiding language bias, however there is no evidence that two independent authors selected the papers and extracted the data.