Sounds interesting, thank you for sharing. There is a secondary analysis in the paper with different assumptions about the effect size, which is a bit more in line with your conclusion (Ext fig 1). That analysis corresponds to a medium effect size per cohen.