I read the article and he basically said there could be as many as 35,000 so it sounds like he was guessing. I don't know how many people the VA is serving but it could be a reasonable estimate given the numbers of people served and prevalence of disease. I think his major point is the VA system needed to know about the test.