Table 3. Logistic regression analysis of current depression and the current and prior status of smoking, depression and drug use.

Four logistic regression models were fitted for the status of depression as outcome variable, with the depression status at the baseline, and the statuses of unhealthy alcohol use, smoking, and one of the four drug uses (1. stimulant use; 2. marijuana use; 3. prescription opioid use; 4. heroin use) at the current and preceding surveys

	Current Depression
	Model 1		Model 2		Model 3		Model 4
Parameter	ORa	p value	ORa	P	ORa	p	ORa	p
Number of years from enrollment	0.98	0.02	0.98	0.02	0.98	0.01	0.97	<.01
HIV status	0.76	<.0001	0.76	<.0001	0.78	<.0001	0.72	<.0001
Depression status at baseline	6.39	<.0001	6.43	<.0001	6.26	<.0001	5.85	<.0001
Current unhealthy alcohol use status	1.13	0.03	1.11	0.06	1.15	0.01	1.11	0.10
Current alcohol use status at preceding surveyb	1.09	0.10	1.09	0.08	1.10	0.06	1.10	0.13
Current smoking status	1.30	<.0001	1.29	<.0001	1.29	<.0001	1.30	<.0001
Current smoking status at preceding surveyc	1.01	0.85	1.01	0.89	1.01	0.91	0.96	0.57
Current stimulant use status	1.83	<.001	-	-	-	-	-	-
Current stimulant use status at preceding surveyd	1.09	0.61	-	-	-	-	-	-
Current marijuana use status	-	-	1.38	<.0001	-	-	-	-
Current marijuana use status at preceding surveye	-	-	0.92	0.17	-	-	-	-
Current opioid use status	-	-	-	-	1.51	<.0001	-	-
Current opioid use status at preceding surveyf	-	-	-	-	1.24	<.01	-	-
Current heroin use status	-	-	-	-	-	-	1.62	<.01
Heroin use status at preceding surveyg	-	-	-	-	-	-	1.03	0.85

‘-‘ indicates that predictor was not included in the model

^aAdjusted for HIV status and number of years from enrollment

^bControlling for current unhealthy alcohol use

^cControlling for current smoking

^dControlling for current stimulant use

^eControlling for current marijuana use

^fControlling for current opioid use

^gControlling for current heroin use