When you submit a sample for genetic analysis you are not just making a decision for yourself.
You are also making a decision for all of your children, their offspring and all of your ancestors, your brothers and your sisters and their offspring.
In fact, you’re making a decision for everybody that is a blood relative.
With the number of bits present in a typical genetic sample this is still more than sufficient to identify people that have not consented (or that might not even exist yet, or any more) to having their genetic data analyzed if you did not submit your sample in an anonymous way. And even that has its limits, after all, your genes are your identity. They identify you with much more precision than any other piece of information. Effectively, you can’t anonymize genetic data at all.
There are many ways in which this could be abused, I can imagine that it could for instance be determined that a disease that runs in the family is an elevated risk, and this could be a reason to deny coverage to sibilings, children or parents of the person tested.
When genetic data from multiple sources is combined it becomes possible to reconstruct the genes of people that have not supplied their data by storing their genetic profiles as probabilities.
For instance, if both parents have supplied a sample then for all of their possible children the probabilities of their genetic make-up are known as well and given a sample you could make statements about whether or not a person is a child of either (or both) parents, or not their child at all. And it would tell you about all the possible genetic defects they might suffer from and allow you to assign probabilities to their genetic make-up further propagating those defects. (But not to the probability of the expression of those defects, that depends on other factors as well, in other words, having a gene for ‘x’ does not guarantee that you will suffer from ‘x’, it merely changes the probabilities).
It doesn’t take a very large number of samples at all before you can start to ‘fill in the blanks’ with an ever higher degree of accuracy. And with every sample submitted data that was mere probabilities before can be nailed down further increasing the accuracy of the remaining blanks.
Another problem here is that having a gene for a certain disease has very little bearing on you personally getting that disease but it has a definite use if you have a population of people that carry that gene, the incidence in the population of the disease will be higher.
To make it a bit more concrete: Say you have a group of people that carry gene that says they are more susceptible to heart disease.
This means that there is a higher chance that those people as a group will develop this disease compared to a population of people that do not carry that disease.
But it does not make any statements with respect to any individual, in other words, people from the group without the variation of the gene may develop heart disease and people from the group with the ‘bad’ gene may not develop heart disease.
These genetic markers are indicators of probability, typically not of certainty (though there are some markers where the probabilities approach certainty, but these are the exceptions and are known to run ‘in the family’, such as for instance a cleft lip).
The only thing that is certain is that when you are tested for lots of diseases is that some of those will come up as more likely than others but this has very little bearing on which diseases you will actually suffer from.
The more you screen the more you will find, but in the end what matters is those diseases that people actually suffer from, not those that you can find increased probability for through some testing method and that includes genetic testing.
Before you submit your data for genetic testing please realize that you are giving away a portion of the ultimate family heirlooms, the genes that run in your family and that this decision could easily come back to bite others.