Problem with the confidence interval of summary statistic in meta-analysis of proportions with subgroups (meta and metafor packages)

Question

Problem with confidence interval of summary proportion in the precense of subgroups. When there is no significant difference between the subgroups, then the 95%CI of the sum