As far as I've been able to tell, nobody gives you a prediction for past five years. Take the low risk one because it was a more established procedure, right? And let them tinker with the numbers for their new study. That's way too small a sample size for you to worry over yet. They just don't have enough to work with.

It's not the same thing at all, but have you followed any of the stories of folks using those family / ancestry DNA kits? They give you some interesting random ethnic information. BUT ... the more of your family try it, the more accurate it supposedly becomes. And if you log in at some future date and find out that more folks with ties to you have also logged in (and this can even be relatives you didn't know you had), it will adjust your percentages accordingly. The more data they have, the more accurate they can be. Wider samples are better. That sets aside the questions I've read about some Native American groups getting mixed up with some Middle Eastern groups (there are discussions on those) and the ethical discussion of what they do with all the data later.

BUT ... it does demonstrate that the larger the sample, the more refined the answers can get. 200 folks is a small sample for something as specific as Oral Cancer risk. I'd do my best to put that out of my head if at all possible (and not even show it to your insurance folks for their perusal, sigh) because your odds are good and that's the part you hold on to.