56% improvement is not statistically significant?
Look at the rest of the data, the tables, and write up. Along with what that footnote applies to and doesn't apply to exactly.
The Confidence Interval for the Cloth Mask data wasn't good enough. So, a 56% reduction based on what they saw, but not great confidence in it.
The Surgical mask and N95/KN95 respirator data had much better modeling results and Confidence Interval.
Which essentially means, wearing a cloth mask will help you protect others from you some, but wearing a Surgical/N95/KN95 will both help protect others from you and protect you from others.
Anyone still wearing just cloth, thank you for offering us some protection. But, really, upgrade your mask to help yourself more too.
I know that I've moved to Surgical for most stuff, and to KN95 for stuff that's extra covidity. The rest of the family mostly too. Unless it's something that's barely covidity at all, but then warmth might be the bigger driver to wearing something.
Alternatively, it's all junk and type of mask wearing is just an indicator for other behaviors, correlation not causation. Maybe all those mask wearer are just doing/not doing something else different. Still good to know, as it's a easy indicator for someone who is doing/not doing whatever that other thing is.