Table 2. Adjusted relative hazards of HIV seroconversion (without accounting for condoms) among North American men who have sex with men from four longitudinal cohort studies, 1995–2007.

				Model with No UAI as reference		Model with No seroadaptive behavior as reference##
Risk category	Visits N (%)	HIV SC* (N)	SC per-cent	Adjusted HR**	95% CI	Adjusted HR	95% CI
No UAI	28,316 (47.6)	173	0.61	reference	n/a	0.31	0.25–0.37
Single negative partner	6393 (10.8)	16	0.25	0.56	0.32–0.96	0.17	0.10–0.30
Exclusive top	6169 (10.4)	25	0.40	0.55	0.36–0.84	0.17	0.11–0.25
Pure serosorting	4437 (7.5)	65	1.44	2.03	1.51–2.73	0.62	0.47–0.82
Pure Seropositioning	2048 (3.4)	15	0.73	0.85	0.50–1.44	0.26	0.15–0.43
No seroadaptivebehavior	12,136 (20.4)	369	2.95	3.27	2.68–3.99	reference	n/a

^*seroconversion.

^**adjusted for age, race, number of sexual partners, any methamphetamine and popper use in the last six months, and intervention assignment.

## models 1 and 2 are equivalent apart from using different groups as the reference category.