Table 2.

Multivariate analysis of MSM’s PrEP willingness and adherence self-efficacy

	Full sample (N = 622) DV: PrEP willingness				PrEP-willing MSM (N = 402) DV: PrEP adherence self-efficacy
	Multivariate logistic regression				Multivariate regression
	B	aOR	95%CI	p	B	95%CI	beta	p
Categorical Variables
Demographic characteristics
Education level
High school or less	0.25	1.29	[0.81, 2.07]	.293
More than high school	ref
Hukou (residence)
Migrant	0.70	2.01	[1.38, 2.92]	< .001***
Local	ref
PrEP awareness
Heard of PrEP					0.32	[−0.16, 0.80]	0.06	.192
No
Yes
Concerns for PrEP usage
The efficacy of PrEP in preventing the transmission of HIV virus					−0.80	[−1.50, −0.09]	−0.10	.027*
Not concerned
Has concern
People may treat me as an HIV/AIDS patient
Not concerned	ref
Has concern	−0.60	0.55	[0.38, 0.80]	.002**
Sexual risk behaviors
Had an HIV+ sex partner in the past six months
No	ref
Yes	1.43	4.19	[1.82, 11.43]	.002**
Sex under influence of substance in past six months
No	ref
Yes	0.94	2.57	[1.67, 4.03]	< .001***
HIV prevention behaviors
Prior use of PrEP	ref
No	1.82	6.17	[1.98, 27.40]	.005**
Yes
Tested for HIV in past 6 months
No	ref
Yes	−0.68	0.50	[0.34, 0.74]	< .001***
Continuous Variables
Sexual orientation concealment	−0.19	0.83	[0.70, 0.96]	.015*
PrEP knowledge					0.34	[0.24, 0.44]	0.32	< .001***
Use of WeChat for HIV prevention	−0.17	0.84	[0.72, 0.98]	.032*

Note. Only variables emerged as significant on the bivariate level (see Table 1) were entered into multivariate regression analysis. Binary variables were dummy-coded (0 = No; 1 = Yes). aOR= adjusted odds ratio; CI = confidence interval;

^*p < .05,

^**p < .01,

^***p < .001