You do when you subset the homeless population from couch surfers and people living in their car to the people actually finding a wink of sleep under some tarps under a noisy overpass
No you don’t. If 50% of society uses drugs, 5% of society is homeless, and 100% of homeless people uses drugs - then you’d see that all homeless people use drugs, but most drug users are not homeless, so it’s not well correlated at all.