How do the companies ensure that their genotyping is accurate and their risk predictions are based on solid genetic and epidemiological statistics?

In this case, the statistics suggest Rwanda may have been lucky: the long-term mental-health effects of the war may not be as widespread as they would have been in people with a different genetic mix.

But so far the only evidence for a genetic component has been through statistics, the relationship being inferred mathematically from comparisons of twins and other such studies of close relatives.