Table 3.

Comparing HIV testing behaviors and sexual behaviors in the past 6 months measured at baseline and follow-up (n = 412).

	Baseline	Follow-up a	P -values
	% Yes	% Yes
Sexual behaviors
Condomless anal intercourse with regular male sex partners	24.0	34.5	0.002
Condomless anal intercourse with non-regular male sex partners	12.4	16.2	0.34
Condomless anal intercourse with male sex workers	2.4	12.9	<0.001
Sexualized drug use	12.1	21.6	0.001
Utilization of HIV prevention services
HIV testing at community-based organization in Shenzhen	11.4	51.6	<0.001
HIV testing at public hospitals or CDC in Shenzhen	20.6	35.2	<0.001
HIV testing at private hospitals in Shenzhen	4.9	13.1	<0.001
HIV testing at other organizations in Shenzhen	9.2	17.2	0.01
HIV testing outside Shenzhen	18.0	23.3	0.16
Home-based HIV self-testing	42.7	52.4	0.001
Any type of HIV testing	59.2	77.9	<0.001

^aMultiple imputation was performed to replace missing values at Month 6. Markov chain Monte Carlo methods were used for data with an arbitrary pattern of missing values, while Monotone methods were used for data having a monotone pattern of missing values. Predictors included baseline background characteristics and baseline value of the variable with missing values at Month 6. The bold values indicate the values of p < 0.05 which are statistically significant.