In most cases the patient's CD4 count is 50 or less, but that is used as more of a "consider prophylaxis" point than a sine qua non - in fact, in at least one study patients with CD4 counts of
less than 100 were just as likely to get CMV disease as those with counts of 50..
I like to think of CD4 cells (and really all immune cells) as little people. You can estimate you need a police force of 50 to keep a certain town under control, but every now and again things get out of hand even with double that amount. It also depends on the quality of the force - I could easily imagine certain HLAs as being more or less susceptible to CMV infection, for instance...