This implementation apparently consumes slightly less processing time than @pajamaclaws21's (excellent!) function - I used [Boolean (unevaluated)], too.
I compared both implementations through the following script:
On my system, running in Turbo mode, @pajamaclaws21's implementation permitted the parallel counter to reach between 200,000 and 250.000, whereas my implementation consistently facilitated 300,000+. It's not a huge difference of course but significant nonetheless.
On the other hand, with my implementation one needs to be careful when specifying the interval, because the waiting thread will keep running until the end of the interval.