Interventions for improving adherence to amblyopia treatments in children.

Rationale: Amblyopia is a neurodevelopmental visual disorder that can be successfully treated, particularly in younger children. However, the effectiveness of treatment is often limited by adherence. This review discusses the efficacy and harms of interventions designed to increase adherence to amblyopia treatments in children.

Objectives: To assess the efficacy and harms of various behavioral interventions for improving children's objective adherence to amblyopia treatments.

Search methods: We searched CENTRAL, MEDLINE, Embase.com, LILACS, ClinicalTrials.gov, and ICTRP. We did not use any date or language restrictions in the electronic searches for trials. The latest search date was 24 May 2024. We also searched the bibliographies of the included trials and any relevant systematic reviews.

Eligibility criteria: We included randomized controlled trials (RCTs) and quasi-RCTs in children aged three to 18 years old at the start of the trial. We included participants with bilateral or unilateral amblyopia, comparing interventions to improve treatment adherence. Interventions were categorized as behavioral (e.g. rewards, informational videos), digital (e.g. video games), and different approaches to patching (e.g. different types of patches). We excluded participants with deprivation amblyopia or with a history of ocular diseases, unrelated to amblyopia, that could affect visual acuity (VA).

Outcomes: Our critical outcome was adherence to treatment using objective adherence measures from one week up to three months. 'Adherence' was defined as the percentage of hours of actual treatment undertaken by the participant compared with the hours of prescribed treatment. Our important outcomes were measured at two time intervals, one week to three months and four to six months: 1. Adherence to prescribed treatment using subjective measures 2. Total hours of adherence 3. Change in recognition VA in the amblyopic eye from baseline 4. Change in random dot stereoacuity We also assessed harms, such as adverse effects and the burden of adherence monitoring, in studies that reported them by the end of the study.

Risk of bias: We used the Cochrane risk of bias (RoB 2) tools for parallel and cross-over trials to assess the risk of bias in the included studies.

Synthesis methods: We synthesized results for each outcome using meta-analyses of mean differences (MD), when data were available, using fixed-effect inverse variance models when fewer than three studies were available and random-effects models if three or more studies were available. Where this was not possible due to the nature of the data, we synthesized the results qualitatively. We used GRADE to assess the certainty of the evidence for prespecified outcomes that were reported.

Included studies: We included a total of 18 studies with 1456 total participants. Eleven studies reported the age of the participants, with mean ages between 4.1 and 6.2 years. Most studies enrolled children with anisometropic, strabismic, and mixed mechanism amblyopia. Four studies reported race and ethnicity, with three of those enrolling predominantly White participants, and one approximately half White and half Other.

Synthesis of results: Behavioral interventions Nine studies compared a single or combination of behavioral interventions to the standard of care or no intervention to improve adherence. Success in improving adherence varied across time frames, with no discernible pattern favoring the use of either a single or a combination of behavioral interventions, or whether the intervention was targeted at children or their caregivers. A majority of studies had a high risk of bias due to missing outcome data or some concerns about the risk of bias due to the described analyses. In general, there was little to no improvement in the change in logMAR VA or random dot stereoacuity at any time point. Digital stimulus interventions Six studies compared digital to non-digital stimulus interventions, with adherence reported as the same or worse with the digital intervention. No study extended past three months. Only one of the five studies evaluating change in logMAR VA reported significant improvement in three months. There was a high risk of bias due to missing outcome data. Only one out of four studies reported improvement in random dot stereoacuity. Different approaches to patching Three studies evaluated different approaches to patching. No study found evidence of change in adherence, VA, or stereoacuity at any time point. Harms No study reported any serious harm from interventions to improve adherence or from the treatment itself (e.g. patching). Non-serious harms included discomfort from occlusion dose monitors; patching expense, rash/itching, inability to tolerate the patch, and negative impact on activities of daily living; reverse or worsening of amblyopia from either patching or digital intervention; and double vision. Limitations of the evidence Studies were mainly excluded due to ineligible comparisons, interventions, or study designs. Several factors reduced data quality and downgraded the evidence certainty, including lack of objective measures of adherence, data heterogeneity, small sample size, imprecision, and non-reporting. Generalizability was limited by participant demographics. Few studies addressed equity-related factors and, when they did, the findings lacked standardization.

Authors' conclusions: Based on our review, the available data were of very low certainty. We were unable to draw conclusions about any differences in adherence, measured objectively, to amblyopia treatments in children using behavioral, digital interventions, or different approaches to patching compared to those receiving no intervention or standard of care. Similarly, we were unable to draw conclusions about any differences in improvement in logMAR VA and stereoacuity arising from these interventions. No data were available for any serious harms, but some data relating to non-serious harms were mentioned.

Funding: This Cochrane review had no dedicated funding.

Registration: Protocol: doi.org/10.1002/14651858.CD015820.