Table 3.

Time and participant type effects in follow-up to MASLIHAT intervention: Mixed effects and population-averaged (GEE) regression models (n=90)

			Contrastsc
Past month alcohol use (# times) a	IRR	95% Conf. Int.	chi2	p-value
6 weeks	0.70	(0.632 – 0.784)	41.15	< 0.001
3 months	0.58	(0.493 – 0.687)	11.39	0.0007
6 months	0.39	(0.297 – 0.508)	19.4	< 0.001
Network member vs. PE	1.01	(0.826 – 1.225)
Any Heavy drinking b	RR	95% Conf. Int.
6 weeks	0.64	(0.422 – 0.959)	4.67	0.0308
3 months	0.42	(0.247 – 0.729)	7.27	0.007
6 months	0.44	(0.257 – 0.742)	1.02	0.313
Network member vs. PE	1.04	(0.635 – 1.692)
Any sex work partner b
6 weeks	0.69	(0.541 – 0.871)	9.62	0.0019
3 months	0.40	(0.275 – 0.572)	13.48	0.0002
6 months	0.20	(0.111 – 0.355)	9.04	0.0026
Network member vs. PE	1.09	(0.737 – 1.611)
Sex without condom b
6 weeks	0.40	(0.251 – 0.637)	14.94	0.0001
3 months	0.10	(0.035 – 0.299)	7.39	0.0066
6 months	0.03	(0.005 – 0.231)	1.84	0.1751
Network member vs. PE	0.85	(0.447 – 1.612)
Injected with used syringe b,d
6 weeks	0.47	(0.313 – 0.700)	13.7	0.0002
3 months	0.04	(0.011 – 0.181)	12.26	0.0005
Network member vs. PE	1.21	(0.789 – 1.849)

^amixed effects Poisson regression

^bGEE Poisson regression with robust variance estimator

^creverse adjacent contrasts for time effects

^d6-month data omitted: 0% injected with used syringe at 6 months

PE: peer educator