Association between digital smart device use and myopia: a systematic review and meta-analysis

Authors: Foreman J, Salim AT, Praveen A, Fonseka D, Ting DSW, He MG, Bourne R, Crowston J, Wong TY, Dirani M.

Geographical coverage: India, Japan, Taiwan, Denmark, Norway, Spain, Singapore, Germany and Ireland.  

Sector: Biomedical

Sub-sector: Risk

Equity focus: Children

Study population: Children and young adults (aged 3 months to 33 years).

Review type: Other review

Quantitative synthesis method: Systematic review and meta-analysis

Qualitative synthesis method: Not applicable

Background: Myopia prevalence is rising globally, with predictions indicating that half of the world’s population will be myopic by 2050. This trend is accompanied by a decrease in the age of onset, an acceleration in progression rate, and an increase in severity at stabilization, potentially leading to a surge in high myopia and its complications, such as irreversible blindness. While the rise in myopia predates the advent of smart devices, there is emerging evidence suggesting these devices could be exacerbating the issue. However, the association has not been thoroughly investigated.

Objectives: To review the literature on the association between digital smart device use and myopia, as excessive use of digital smart devices, including smartphones and tablet computers, could be a risk factor for myopia.

Main findings:

A total 33 studies were included in the systematic review and 11 in the meta-analysis. The review identified several sources of bias in the 33 studies it examined. These include the lack of valid or reliable measurement of exposure, absence of objective standard criteria for measuring the condition, and insufficient strategies for dealing with confounders. Most of the studies investigated Asian populations, with some focusing on European populations, but none from other world regions. The majority of the studies were cross-sectional, with a few being prospective. Population-based surveys, such as the North India Myopia (NIM) study, used cluster sampling of classes, schools, or districts.

In four out of ten articles where smart devices were examined as an independent risk factor (category one), eight out of ten articles where smart devices were investigated alongside computer exposure, and all thirteen category three articles (those where smart device use was studied, but not independently from other near-vision activities), objective measures were used to identify myopia. The remaining studies utilized questionnaires for this purpose. All studies measured screen exposure using questionnaires, with one study also measuring device-recorded network data consumption.

Five (50%) category one and six (60%) category two articles reported associations between screen exposure and prevalent or incident myopia, an increased myopic spherical equivalent, and longer axial length. Myopia was significantly associated with smart device screen time alone (OR 1.26 [95% CI 1.00-1.60]; I² = 77%) or in combination with computer use (1.77 [1.28-2.45]; I² = 87%).

The most common sources of bias were the lack of reliable measures of screen time in all 33 studies; seven (21%) did not objectively measure myopia; and nine (27%) did not identify or adjust for confounders in the analysis. High heterogeneity between studies in the meta-analysis was due to variability in sample size (range 155-19,934 participants), mean age of participants (3-16 years), standard error of the estimated odds of prevalent or incident myopia (0.02-2.21), and the use of continuous (six [55%] of 11) versus categorical (five [46%]) screen time variables. Smart device exposure might be associated with an increased risk of myopia. Research with objective measures of screen time and myopia-related outcomes that investigates smart device exposure as an independent risk factor is required.

Methodology:

Included studies in this review investigated myopia-related outcomes of prevalent or incident myopia, myopia progression rate, axial length or spherical equivalent. Studies were excluded if they were reviews or case reports, did not investigate myopia-related outcomes, or did not investigate risk factors for myopia.

The authors conducted a comprehensive search of Medline and EMBASE, as well as manual searches of reference lists, to find primary research articles investigating the relationship between smart device usage and myopia in children and young adults aged between 3 months and 33 years. The search was carried out from the inception of the databases until June 2nd (Medline) and June 3rd (EMBASE) 2020. The search terms used were broad enough to include publications that considered smart devices as one of many risk factors for myopia. The search was not restricted by language.

Two reviewers screened all titles and abstracts. Both reviewers read the full-texts of all remaining articles. Articles were excluded if risk factor analysis did not include mobile phones or tablets, either separately or combined with other forms of near-vision tasks, or if myopia-related outcomes were not measured. Conflicts over inclusion were adjudicated by a third reviewer.

Three authors extracted data from the studies, focusing on variables such as study design, sampling methodology, sample size, participants’ age and country of residence, response rates, and the definition and measurement of myopia. They also considered screen exposure measures, myopia-related outcomes, and statistical associations between smart device exposure and myopia-related outcomes. Additionally, they looked at variables that were adjusted for in multivariable analysis of associations between smart device screen exposure and myopia-related outcomes.

All articles were appraised using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Analytical Cross-Sectional Studies and the JBI Critical Appraisal Checklist for Cohort Studies to assess their methodological quality and risk of bias. Studies with unclear statistical analysis or reporting of results were excluded.

The meta-analysis included studies that reported adjusted odds ratios (ORs) or other adjustable measures, like β coefficients, that could be converted to ORs. These measures were used to examine the association between smart device exposure and prevalent or incident myopia. The studies were categorized into three groups: category one studies investigated smart devices as an independent risk factor; category two studies examined smart devices, but not independently of computer screen exposure; and category three studies investigated smart device use, but not independently of other near-vision activities like watching TV, reading non-digital books, and writing.

A meta-analysis of the association between screen time and prevalent or incident myopia for category one articles alone and for category one and two articles combined was undertaken by the authors. Random-effects models were used when the study heterogeneity was high (I² >50%) and fixed-effects models were used when heterogeneity was low (I² ≤50%).

Applicability/external validity:

The majority of the studies were conducted in Asian populations, making it uncertain whether the results can be applied universally. Less than a third of the studies differentiated smart device screen time from other near-vision activities, and due to inter-study heterogeneity, multiple meta-analysis models were required, making it difficult to draw firm conclusions about the relationship between smart device usage and myopia. Furthermore, all but one study (McCrann et al.) relied on self-reporting or parental-reporting to measure digital screen exposure, limiting the accuracy of the data.

Geographic focus: The studies included were carried out in a variety of high-income and low-to-middle-income countries. However, it’s important to note that the majority of these studies were conducted in Asian populations.

Summary of quality assessment:

The methods used to identify, select, and critically evaluate the studies were thorough, with multiple authors involved in all crucial tasks and careful consideration given to the risk of bias in individual studies. However, there was no mention of efforts to include unpublished materials or to reach out to relevant experts. The analysis of the included data was generally robust, but there was no attempt to analyze subgroups of studies based on their varying levels of bias. Due to these factors, we have medium confidence in the review’s findings.