Figure 3 of
Ayala-Lugo, Mol Vis 2007;
**13**:151-163.

Figure 3. Studywise differences appear in Japanese populations when odds ratios and frequencies of M98K mutations are compared

The left-hand graph (**A**) shows the odds ratios with 95%
confidence interval bars for individual Asian studies and pooled results
for Japan, China, and both in open angle glaucoma (OAG) cases versus
controls. Odds ratios and confidence intervals are fixed effect
estimates resulting from the Mantel-Haenszel method. The right-hand
graph (**B**) shows the case (OAG, filled circle) and control (open
circle) proportions observed for each study. Total sample sizes are
listed along the right-hand margin. Larger samples have both narrower
confidence intervals and shorter distance between fractions observed for
cases and controls. Studies inconsistently estimate the odds of OAG
versus controls carrying an M98K mutation, with larger studies (more
than 400 total cases and controls) estimating no statistically
significant difference. Other population estimates are not shown,
because, among the European population-based studies, only Rezaie's
study [9] showed a statistically significant difference. The single
study on India yielded a significant odds ratio, but no other comparable
populations have been reported [19].

Ayala-Lugo, Mol Vis 2007;

©2007 Molecular Vision <http://www.molvis.org/molvis/>

ISSN 1090-0535