Table 2.

Univariable and Multivariable Logistic Regression Models for 1-1 Match Case-Control to Predict HIV Infection, n = 128 Pairs, Male Injection Drug Users, Bac Ninh Province, Vietnam.

Effect	Univariable Models			Multivariable Model*
	OR	95% CI	p-value	OR	95% CI	p-value
Length of drug injection (yrs)**	1.07	0.98–1.16	0.146	1.06	0.96–1.16	0.267
Age at first drug injection (yrs)**	1.01	0.93–1.09	0.856
Injecting drugs with needles previously used by another, past 6 months	0.86	0.40–1.85	0.696
Sharing water for mixing drugs or cleaning works, last injection	10.00	1.28–78.11	0.028	4.63	0.53–40.19	0.164
Sharing drug solution through frontloading, last injection	3.22	1.53–6.81	0.002	2.75	1.17–6.48	0.021
No. of sexual partners, past 12 months	0.92	0.84–1.01	0.094	0.94	0.84–1.05	0.241
Having had sex with a female sex worker, past 12 months	0.68	0.42–1.13	0.137	0.72	0.37–1.43	0.350

OR denotes odds ratio; CI, confidence interval;

^*Multivariable model included Length of drug injection, Sharing water for mixing drugs or cleaning works at last injection, Sharing drug solution through frontloading at last injection, Number of sexual partners in the past 12 months, and Having had sex with a female sex worker in the past 12 months;

^**Odds ratio for one year increase.