Fig. 3.

Summary risk estimates reported for the association between NSAID use and proximal colon cancer by subgroup. ^aEstimates were log-transformed prior to generation of the pooled estimate; ^bExcluding Cheung 2020 [67] due to different outcome measure, i.e., post-colonoscopy proximal colon cancer within 3 years; ^cExcluding Allison 2006 [59] for which no association was found but no risk estimate was reported. CI Confidence interval, I² test for heterogeneity between studies within subgroups, Meta-regression p-value p-value for evidence of between subgroup heterogeneity, NSAID Non-steroidal anti-inflammatory drug