> how these studies control for confounding factors like that

The answer is: they simply do not at all.

[1] http://www.salem-news.com/fms/pdf/2011-12_JLM-Boyle-Hill.pdf

[2] http://www.circumstitions.com/HIV-SA.html