> potentially having to rely on a result that may or may not have statistical significance
I can't stress this enough. I've had a bunch of friends buy the 23andme kit while it was on sale recently because they were interested in my results from a while ago, and their buildup to getting their own results was "what if it tells me I'm never going to have x?!" and things like that. Kind of frustrating to explain how it works.
23andme put me at average to low risk of a few things they test for that I have in real life. One of the more frustrating things to read was that 23andme declared from preliminary asthma research that I was at lower than average to average odds, when in my past I've had severe asthma I've even been hospitalized for.
On the flip side of things, 23andme determined my risk of melanoma from genetics is 0.1% (as compared to 1.7% average). But I live in Los Angeles and I get sunburnt/suntanned multiple times a year, not to mention the sheer number of hours I spend outdoors in general. I regularly track skin changes and visit a dermatologist once in a while because new spots and changes on my skin stress the hell out of me. I know my risk likely skyrockets past the 1.7% easily if you take my environment into account.
Insurance companies already play this risk/percentage game anyway with lots of other information, and I don't think I see a big of a risk as people being denied insurance from testing like this. If anything, it means everyone can be more proactive and save money and lives in the end. (Of course, I am being pretty naively optimistic here...) I mean, I imagine that's what they're trying to do on some level when they send me pamphlets about STD testing and prenatal care in the mail every other week just because they know I'm a 20something female on birth control.