Degree of myopia and glaucoma risk: a dose-response meta-analysis

Authors: Ha A, Kim CY, Shim SR, Chang IB, Kim YK

Geographical coverage: India, China, Myanmar and Sri Lanka

Sector: Burden of disease

Sub-sector: Epidemiology

Equity focus: None

Study population: General population

Review type: Other review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis method: Not applicable

Background: Myopia is a well-established risk factor for open-angle glaucoma (OAG), the association between them having been thoroughly investigated. The evidence on the association of myopia degree with increased risk of OAG, however, is contradictory. According to some studies, an association exists between myopia of any degree and OAG, whereas other investigations have reported links only with high myopia. A 2011 study analysed 13 population-based studies and reported that myopic individuals are at increased risk of OAG and that the odds of developing that disease are slightly increased in myopia of higher degrees; however, authors had classified myopia into two groups and included studies, individuals and ethnicities were relatively small in number.

Objectives: To verify the dose-response relation between the degree of myopia and open-angle glaucoma (OAG) risk.

Main findings:

Overall, authors found that for each unit (1-D) increase in myopia, the risk of glaucoma increases by approximately 20%. The risk more steeply increases in high-degree myopia, representing a significant nonlinear relationship.

Authors included 24 studies in the review. The median score on the Newcastle-Ottawa Scale was 8 of 9, with a minimum of 6 and a maximum of 9. The 24 studies, with a total study population of 514,265 individuals, had been conducted in 11 countries: four studies in the United States, four in India, three in China, two in Japan, two in South Korea, two in Australia, three in Singapore (Malay, Indian and Chinese), and one study each in Myanmar, Sri Lanka, Greece and Barbados. In 13 studies, myopia was defined based on an SE of -0.5 D, and seven others reported myopia of SE less than -1.0 D.

The meta-analysis comprised 24 studies in 11 countries (514,265 individuals). The pooled OR of any myopia degree’s association with OAG was 1.88 (95% CI, 1.66-2.13; I² = 53%). The OR differences based on ethnicity (Asians vs Westerners) or five geographic areas were not statistically significant (p = .80 and p = .06, respectively). The pooled ORs of the associations between low, moderate, moderate-to-high, high myopia, and OAG were 1.50 (95% CI, 1.29-1.76), 1.69 (95% CI, 1.33-2.15), 2.27 (95% CI, 1.74-2.96), and 4.14 (95% CI, 2.57-6.69), respectively. According to the dose-response meta-analysis, the pooled OR (per SE 1-D change) was 1.21 (95% CI, 1.15-1.28). The OAG risk accelerated at approximately -6 D, and further accelerated from -8 D, showing a nonlinear concave upward slope (p = .03).

In terms of subgroup analysis, the OR difference between studies including Asians (OR, 1.945; 95% CI, 1.570-2.409) and Westerners (OR, 1.868; 95% CI, 1.497-2.331) did not reflect any statistical significance (Q = 0.07, p = .80). The subgroup analysis according to geographic area (ie, five continents) showed relatively higher ORs for Australia (OR, 2.220; 95% CI 1.429-3.449), Northeast Asia (OR, 2.190; 95% CI, 1.703-2.816), and Southeast Asia (OR, 2.172; 95% CI, 1.197-3.944) compared with South Asia (OR, 1.681; 95% CI, 0.971-2.909) or North America (OR, 1.601; 95% CI, 1.549-1.655), with borderline significance (Q = 8.87, p = .06).

Authors note that future studies performing ideal structural evaluation of myopic eyes should consider refractive error and AXL along with features of the ONH.

Methodology:

Inclusion criteria consisted of (1) population-based study; (2) myopia reported as covariate; (3) OAG as outcome measure; and (4) measure of association reported as odds ratios (ORs) with 95% CIs, or the allowed calculation from the data presented in the article. The exclusion criteria were (1) not conducted with humans or adults; (2) narrative and/or systematic reviews, commentaries, case reports; (3) involving secondary glaucoma or angle-closure glaucoma; and (4) lacking detailed definition of OAG or without clear description of myopia assessment. In situations where multiple publications were available for the same study population, only the study with the largest cohort was included.

The authors conducted a comprehensive search of the PubMed, EMBASE, and Cochrane Library databases for relevant studies. Two investigators independently and blindly performed the search, with any discrepancies resolved through discussion or, if necessary, by a third investigator1. Reference lists of retrieved articles were manually reviewed to identify additional relevant studies. The search included all relevant reports published up to 30 November 2020. Data extraction was performed independently and blindly by two investigators using a standardised method2. The Newcastle-Ottawa Scale (NOS) was used to assess the methodological quality of comparative nonrandomised studies. Furthermore, studies were evaluated for potential selection, comparability, exposure/outcome biases, or any other forms of bias.

Most studies reported odds ratios (ORs) for any myopia and stratified myopias. Stratified ORs were pooled for an overall estimate for any myopia. The pooled OR for 1-mm increments in axial length (AXL) and open-angle glaucoma (OAG) risk were analyzed. Fully adjusted, study-specific ORs were combined to estimate the pooled OR with a 95% confidence interval (CI) using a random-effects model. Interstudy heterogeneity was quantified using the I2 statistic. Influence analyses were performed to determine the influence of any study on the overall results. A graphic display of heterogeneity (GOSH) plot analysis was used, which employs three clustering algorithms. Sensitivity analysis was used to determine the effect of rerunning the meta-analysis after removing potential outlier studies.

In subgroup analyses, authors examined subgroups differing by ethnicity and geographic area. For studies reporting categorised ORs according to myopia degree, a pooled OR was calculated for each of the four different myopia groups. A dose-response meta-analysis (DRMA) was performed to confirm the dose-response relationship between myopia and OAG risk. Publication bias was evaluated both qualitatively and quantitatively. All statistical analyses were performed with R 4.0.4 software, except for the 2-stage random-effects DRMA, which was run using Stata 14.2 software.

Applicability/external validity: Authors indicated a number of limitations, including the use of different diagnostic approaches between studies, which may limit the applicability of the reported results.

Geographic focus: A number of the included studies were conducted in low- and middle-income countries. However, authors do not consider how reported findings might vary in different regions.

Summary of quality assessment:

The methods used to identify, include and critically appraise studies were generally robust, with all key tasks being undertaken by at least two authors. However, there is no evidence of attempts being made to include unpublished material in the review. While the approach to the analysis of the data, via a network analysis, was generally robust, it is not clear that the quality of included studies was considered as a specific variable in the analysis, with no separate subgroup analysis being undertaken on this basis. For these reasons, we have attributed a medium confidence in the findings of this review.