In the name of brevity: The problem with binary HIV risk categories