so i'm going to do some math, related to genetic tests. and i'm going to use some really simple numbers.
suppose there's a disease called "purple wart disease" that has a genetic component: those who have the PPE gene have an 80% chance of developing the disease and those who lack the PPE gene have only a 10% chance. Luckily, only 10% of the population have the PPE gene.
So, for the population as a whole, if one =doesn't= know whether or not one has the gene, the probability of getting purple wart disease is 17%. (=90%*10% + 10%*80%)
Suppose there's a treatment for purple wart disease that costs $1000. So, the expected amount, per person, to be spent on the disease is $170 for the general population. An insurance company develops an insurance policy for purple wart treatment, where this payment would be covered, and charges $190/person (they need to profit, after all).
Things are fine.
Then a test for the PPE gene comes out. But by law, insurance companies cannot find out the info. The test is cheap (and the purple wart disease is really tacky), so everyone who holds a purple wart policy gets the test.
What happens? For those who =don't= have the disease, the expected amount to be paid on purple wart disease is only $100 ($1000*10%) -- the $190 price for the policy is just too high. So they drop their policies. Those with the PPE gene have an expected payment of $800 ($1000*80%), so the $190 price is a great deal. People with the PPE gene buy up the policies in great amounts...
However, the actuaries could see this coming, and have now priced the policies at $880. Though they take a loss on some of the grandfathered policies, this is offset by the policies paid for by people without the PPE gene.
However, you end up with some unhappy people. Even people without the PPE gene would be willing to pay for purple wart insurance - if it were priced right for their situation. Most people would like to have a zero risk situation - where they pay some amount no matter what - rather than get surprised and have to pay a large amount, even if it's at a small probability.
I think you can see the applicability tp genetic testing now - though this is an =extremely= simplified example. one way to avoid this situation is to have information available to both insurance companies and clients, and pricing can be based on individual info -- however, this is unpalatable to most people, as the pricing for people with "bad genes" would be priced beyond their means...and it doesn't seem very fair, as people can't do anything about their genes (for now).
Another way to deal with this is for the government to cover people of high risk (hmmm, this reminds me of terrorism insurance =cough=).
And yet another way to deal with this is to force everybody to buy insurance - so the expected payout by the insurance company is $170/person. Again, people will be pissed off because those who know they don't have the PPE gene will know they're paying too much and feel like they're subsidizing those with the PPE gene.
However, the individual prices are dependent on the actual numbers. If the prevalence of the PPE gene were much less, then the expected payout would drop. If only 1% had the gene, then the expected payout per person for the population would be: $107. This doesn't differ terribly from the expected payout for a non-PPE person: $100.
In any case, as genetic tests aren't terribly cheap, their use isn't very widespread, and thus have little impact for now. There is info the insurance companies can use to price policies, like age, which have an even larger impact on prices (older people are more likely to be sick, and have more costly treatments). Still, an imbalance in information can cause problems in private insurance.