Table 3.

Multivariable logistic regression of the association between MOS-SSS and experiencing ≥1 TI, controlling for confounders

Variables	Univariable Logistic Regressions		Multivariable Logistic Regression
	Odds ratio (95% CI)	P-value	Adjusted odds ratio (95% CI)	P-value
MOS-SSS score (per ten unit increase)b	0.92 (0.85, 0.99)	0.031	0.90 (0.83, 0.99)	0.022
Age at interview
Less than 40	Ref
40 to 49	0.70 (0.40, 1.22)	0.208	Not selecteda
50 to 59	0.41 (0.23, 0.74)	0.003
60 or more	0.25 (0.11, 0.58)	0.001
Gender
Male	Ref		Ref
Female	2.58 (1.61, 4.15)	< 0.001	1.75 (1.01, 3.03)	0.045
Other	1.51 (0.32, 7.15)	0.605	1.46 (0.27, 7.81)	0.659
Ethnicity
Caucasian	Ref		Ref
Indigenous	3.27 (1.93, 5.53)	< 0.001	2.46 (1.35, 4.49)	0.003
Asian/African/Black/ Caribbean/Latin	0.65 (0.23, 1.90)	0.435	0.83 (0.28, 2.48)	0.744
Other	2.02 (0.94, 4.32)	0.070	1.82 (0.81, 4.08)	0.145
Employment status
Working	Ref
Unemployed	2.50 (1.49, 4.20)	0.001	Not selecteda
Other	1.10 (0.57, 2.13)	0.780
Incarceration ever
No Yes	Ref 3.26 (2.09, 5.09)	< 0.001	Not selecteda
Homelessness in the last 12 months
No Yes	Ref 3.24 (1.92, 5.49)	< 0.001	Ref 2.21 (1.25, 3.90)	0.006
Mental health diagnosis ever
No Yes	Ref 1.01 (0.64, 1.61)	0.958	Not selecteda
Sexual activity in the last 12 months (n = 590)
No Yes	Ref 0.98 (0.61, 1.55)	0.913	Ref 1.49 (0.89, 2.50)	0.131
History of IDU
Never	Ref		Ref
Yes but more than a year ago	2.05 (1.18, 3.57)	0.011	1.78 (0.98, 3.23)	0.060
Yes in the past year	3.72 (2.21, 6.26)	< 0.001	2.53 (1.43, 4.50)	0.002
Years of follow-up	1.14 (0.87, 1.51)	0.351	1.07 (0.80, 1.43)	0.666

Bold text refers to p-values < 0.05

CI: confidence interval

IDU: injection drug use

^aConfounders were selected for inclusion in the final model using a backward stepwise selection approach, which used the relative change in coefficients of the MOS-SSS score as a criterion, until the minimum change from the full model exceeded 5%

^bMOS-SSS odds ratio and adjusted odds ratio reported per 10 unit increase. Reference category for TI: ‘No’