CMV, Serologic Testing, and Leukoreduction

For years, blood bankers and physicians have debated the relative merits of preventing CMV infection through the exclusive use of blood from donors who do not have demonstrable anti-CMV antibodies vs. blood that is leukocyte reduced. The data is mixed, and the two methods seem to be nearly equivalent (and imperfect). Why both fail at roughly the same low rate (about 1-4%) has never been particularly clear to me, until recently.