Prevalence Ratio of Primary Angle-Closure and Primary Open-Angle Glaucoma in Asian Population: A Meta-Analysis and Multiple Meta-Regression Analysis

Ethics statement

This study was conducted under the guidance of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [9]. Research that consist of a meta-analysis are exempt from the need for Institutional Review Board approval [10].

Search strategy and study selection

The PubMed, Embase, and Cochrane search engines were referenced in order to identify the relevant studies. The search items selected were based on an established terminology that included the MeSH (Medical Subject Headings) and Embase extensive search terms whenever possible. The keywords were as follows: glaucoma, open-angle glaucoma, angle-closure glaucoma, and prevalence. Two reviewers (JL, JSP) independently screened the titles and abstracts. The full texts were sourced for the relevant articles. The inclusion criteria were independently assessed, and any inconsistencies were resolved via consensus. We also manually reviewed the retrieved articles’ reference lists to identify additional relevant studies. The databases were searched for reports that had been published through August 5, 2022. The flowchart which summarizes the study selection process is shown in Supplementary Fig. 1.

Our aim was to identify all population-based studies that had reported the prevalence of both POAG and PACG simultaneously in Asian adults aged 18 years or older. Excluded were the following studies: (1) studies not conducted with humans or adults; (2) narrative and/or systematic reviews, commentaries, abstracts, and case reports; (3) lacking any detailed glaucoma assessment explanation; and (4) studies not conducted in Asia. In cases where more than one study had been published by a single institution, we confirmed their independence and absence of overlap in their research subjects or methods.

Data extraction and quality assessment

Two investigators (YJ and YIS) extracted data in an independent and masked manner using a standardized method of data extraction based on the ones employed by the Cochrane Database of Systematic Reviews (Review Manager ver. 4.1.1). The following study data were extracted: (1) name of first author; (2) year of publication; (3) race/ethnicity of study population; (4) country of study; (5) number of subjects; (6) ages and sexes of participants; (7) number of patients with POAG; and (8) number of patients with PACG.

Statistical analysis

We conducted a meta-analysis for prevalence ratio of PACG and POAG using inverse variance–weighted random-effects meta-analyses so as to combine the study-specific measures of association. Between-study outcome variation (i.e., heterogeneity) was quantified with the I² statistic, which describes the percentage variation across studies due to heterogeneity rather than to chance, regardless of the treatment effect metric [11].

In subgroup analysis by the random-effects model for between-subgroup differences, we looked at subgroups differing by the following: (1) number of study population; (2) baseline mean age; (3) prevalence of POAG; (4) continents; and (5) publication year.

The meta-regression analysis entailing the random-effects model were performed in order to further account for the reasons for heterogeneity [12]. Each moderator was included in the random-effects univariate meta-regression analysis by maximum likelihood estimation. The covariates included were as follows: (1) study population; (2) baseline mean age; (3) prevalence of POAG; (4) continents; and (5) publication year. In multiple meta-regression [13], we modeled with all possible predictor combinations to examine the possible predictor combination which provides the best fit and identify the most important predictor.

Publication bias was graphically evaluated with funnel plots and statistically evaluated with Egger linear regression method test [14,15]. This test determines if there is any significant correlation between effect estimates and their variances. The absence of such correlation suggests that studies were selected in an unbiased way.

Data handling and the statistical analyses were performed by one independent data analyst (MGH). The meta-analysis was performed with R ver. 4.1.4 (R Foundation for Statistical Computing). A p < 0.05 were considered to be statistically significant. All p-values and 95% confidence intervals (CIs) were two-sided.

Search results and study characteristics

The search of articles published before August 5, 2022, returned 973 results. After the title and abstract screening, 47 articles were found to be potentially relevant. Following the thorough full-text review, 20 studies were selected (Table 1) [16–39]. All studies were of population-based design, had originated from Asian countries.

Table 1

Characteristics of studies included in the meta-analysis

The 20 studies, with a total study population of 52,522 individuals, had been conducted in 13 countries: five studies in China [16–21], three in India [22–25], two in South Korea [26–28], and one study each in Japan [29, 30], Myanmar [31], Singapore [32], Sri Lanka [33], Bangladesh [34], Thailand [35], Nepal [36], Mongolia [37] , Iran [38], and Laos [39]. The characteristics of all of the included studies are summarized in Table 1 [16–39].

Overall summary measurement

The pooled PACG to POAG prevalence ratio was 2.204 (95% CI, 1.617–3.004) with high heterogeneity (I² = 100%, p < 0.001). Fig. 1 and 2 provide the relevant forest plot (a graphical display of the observed effect and 95% CI) and drapery plot (a confidence curve is plotted for each study, as well as for the average effect) [16–39].

Fig. 1

Forest plot of primary angle-closure glaucoma to primary open-angle glaucoma prevalence ratio in Asian population. The diamond denotes the pooled ratio, and the lateral tips of the diamond indicate the associated confidence intervals (CIs).

Fig. 2

Drapery plot of primary angle-closure glaucoma to primary open-angle glaucoma prevalence ratio in Asian population. The resulting plot contains a p-value curve for each effect size, all in the shape of an upside-down V. The thick line represents the average effect according to the random-effects model. The shaded area we see in the plot represents the prediction interval, which is considerably wider than the confidence interval of the pooled effect. The “peak” of the p-value functions represents the exact value of the effect size in our meta-analysis.

Subgroup analysis

We performed subgroup analyses to investigate the potential sources (i.e., number of study population, baseline mean age, prevalence of POAG, continents, and publication year) of between-study heterogeneity. The subgroup analysis according to prevalence of POAG (i.e., <2.0% vs. ≥2.0%) showed relatively higher PACG to POAG prevalence ratio for ≥2.0% (3.00; 95% CI, 2.106–4.270) compared with <2.0% (1.39; 95% CI, 0.930–2.075), with statistical significance (Q = 7.96, p < 0.001). However, the PACG to POAG prevalence ratio between studies did not reflect any statistical significance among subgroups by number of study population, baseline mean age, continents, and publication year (all p > 0.01).

Univariate meta-regression analysis

The subgroup analyses showed a possible modulating influence of prevalence of POAG. Thus, we performed a random-effects meta-regression analysis to see that the estimate of the heterogeneity variance that is explained by the predictor (i.e., prevalence of POAG). The model’s estimate of the regression weight for elevation of POAG prevalence is 0.471 (Fig. 3). This means that for every additional elevation of POAG prevalence (i.e., increase of 1.0%), the effect size of a study is expected to rise by 0.471. Therefore, we can say that the POAG prevalence of studies have a positive association with PACG to POAG prevalence ratio. The 95% CI ranges from 0.191 to 0.751, showing that the effect is significant (p < 0.001).

Fig. 3

Bubble plot for visualizing a meta-regression. The resulting plot shows the estimated regression slope, as well as the effect size of each study (i.e., primary angle-closure glaucoma [PACG] to primary open-angle glaucoma [POAG] prevalence ratio) according to the covariate (i.e., prevalence of POAG). To indicate the weight of a study, the bubbles have different sizes, with a greater size representing a higher weight.

Multiple meta-regression analysis

In multiple meta-regression model, possible predictor combination provides the best fit was prevalence of POAG and continent (corrected Akaike information criterion [AIC], 37.3). In Fig. 4, the averaged importance of each predictor across all models is displayed. Prevalence of POAG (model importance, 0.954) is the most important predictor, followed continent (model importance, 0.508) and publication year (model importance, 0.222).

Fig. 4

Model-averaged predictor importance plot. The averaged importance of each predictor across all models is displayed. POAG = primary open-angle glaucoma.

Publication bias

Fig. 5 is the funnel plot depicting publication bias. No publication bias was evident in this visual examination of plot symmetry. Likewise, there was no evidence of publication bias by the Egger test (p = 0.321).

Fig. 5

Funnel plots of publication bias regarding primary angle-closure glaucoma to primary open-angle glaucoma prevalence ratio. The middle-dotted line indicates the meta-analysis’s overall effect, while the two dotted lines on either side represent the 95% confidence intervals.