Table 4. Effect of crowdsourced intervention on secondary outcomes among Chinese MSM, 2016–2017: Generalized linear mixed models (N = 1,219).

Secondary outcomes¶	Estimated risk ratio (95% CI)*	p-value	ICC by city
HIV self-testing	1.89 (1.50, 2.38)	<0.001	0.028
HIV facility-based testing	1.00 (0.79, 1.26)	0.99	0.002
Condom use	1.00 (0.86, 1.17)	0.96	0.007
Syphilis testing	0.92 (0.70, 1.21)	0.55	0.005
Using Weibo to give/receive information#	0.95 (0.77, 1.19)	0.66	0.010
Using WeChat to give/receive information#	1.18 (0.51, 2.75)	0.24	<0.001
Using QQ to give/receive information#	0.88 (0.71, 1.09)	0.25	0.030
Using gay mobile phone apps to give/receive information#	0.95 (0.35, 2.56)	0.62	<0.001
Increased community engagement£	0.97 (0.44, 2.12)	0.70	<0.001
	Estimate (95% CI)*	p-value
	Mean difference
Anticipated HIV stigma	−0.027 (−0.064, 0.010)	0.15	0.006
HIV testing social norms	−0.010 (−0.041, 0.020)	0.51	0.002
HIV testing self-efficacy	−0.008 (−0.039, 0.023)	0.62	<0.001

*Adjusted for secular time trend as a fixed effect across clusters.

^#Using WeChat, Weibo, QQ, or Blued to give or receive information about HIV testing, excluding receipt of intervention materials.

^¶We included 1,219 participants who filled out at least one of the four follow-up surveys in this analysis.

^£Defined as whether the cumulative community engagement score increased, by comparing to the baseline.

Abbreviations: CI, confidence interval; ICC, intraclass correlation coefficient; MSM, men who have sex with men.