Exactly. This sounds like selection bias in action. If you're sampling adults who have ADHD and asking if it's better from when they were kids, you've already skewed the results beyond repair. You need to follow kids with ADHD into adulthood and see if it gets better.