I work with NIED as well—the group you are talking about—on these studies. It is accurate. The challenge is that there isn't great death data for a lot of these indicators, so it's really hard to validate that.
It's hard to ascertain exactly how many people have an eating disorder. There isn't a really great tracker for it, so we use a risk screener, the ODES-Y, to ascertain the potential risk for it. It's not a great screener, to be honest with you, but I have seen the numbers you're quoting. I have talked with them about that, and yes, they do seem correct to me.