Whilst everything else of your post is on point, the last bit is not really applicable: you can’t really compare “lifetime” probability (i.e. for the age range 0 - life-expectancy) of getting something (i.e. “be affected by”) with the probability of actually having something (i.e. not just be affected by it at any one point but rather being now suffering the effects from it) whilst being in a specific age range (roughly 10 - 30, a subset of the lifetime one).
It might be possible to derive the second one from the first if knowing the statistic distribution in relation to age of that disorder and the average duration of the condition, but as it stands there those 38% aren’t comparable to those 33.7% as they’re statistically quite different things.