Skip to main content
. 2014 Mar 6;9(3):e90959. doi: 10.1371/journal.pone.0090959

Table 2. Multiple logistic regression analysisa of factors associated with depression and suicidal ideation among participants (N = 292b).

Depression (BDI-Ia≥20) Suicidal ideation (BDI-Ia item #9>0)
Variable AOR (95% CI) AOR (95% CI)
Gender
Female 0.61 (0.25, 1.48) 0.40 (0.14, 1.13)
Male (Ref) 1.00 1.00
Age (years)
34–60 2.01 (0.95, 4.25) 0.62 (0.26, 1.51)
0–33 (Ref) 1.00 1.00
Current marital status
Married 1.43 (0.64, 3.17) 0.82 (0.33, 2.07)
Unmarried (Ref) 1.00 1.00
Any children
Yes 1.24 (0.55, 2.78) 0.85 (0.33, 2.22)
No (Ref) 1.00 1.00
Education level
Secondary or higher 0.68 (0.33, 1.40) 0.54 (0.23, 1.27)
Primary or lower (Ref) 1.00 1.00
Employment status
Employed 0.46 (0.22, 0.99)* 0.27 (0.12, 0.63)**
Unemployed (Ref) 1.00 1.00
Time since first testing HIV+, months
54–258 1.07 (0.53, 2.19) 0.88 (0.38, 2.08)
0–53 (Ref) 1.00 1.00
Months on ART
25–120 1.85 (0.63, 5.46) 2.92 (0.77, 11.11)
0–24 2.92 (1.13, 7.51)* 3.61 (1.15, 11.39)*
Not currently on ART (Ref) 1.00 1.00
Underweight (BMI<18.5)
Yes 3.29 (1.06, 10.20)* 1.21 (0.34, 4.30)
No (Ref) 1.00 1.00
Any drug use, last 6 mos.
Yes 2.86 (1.07, 7.66)* 3.63 (1.23, 10.73)*
No (Ref) 1.00 1.00
Bothersome HIV symptom count
3–16 4.28 (2.09, 8.80)** 3.71 (1.54, 8.94)**
0–2 (Ref) 1.00 1.00
Internalized AIDS stigma score
5–7 3.37 (1.66, 6.87)** 1.81 (0.77, 4.24)
0–4 (Ref) 1.00 1.00
Disclosure of HIV status to any family members
Yes 2.05 (0.80, 5.23) 1.56 (0.52, 4.64)
No (Ref) 1.00 1.00
Perceived family support
High (Total score: 27–30) 0.19 (0.07, 0.55)** 0.25 (0.07, 0.91)*
Moderate (Total score: 23–26) 0.38 (0.17, 0.86)* 0.46 (0.17, 1.19)
Low (Total score: 0–22) (Ref) 1.00 1.00

AOR, adjusted odds ratio. CI, confidence interval. ART, antiretroviral therapy; BDI, Beck Depression Inventory; BMI, body mass index.

a

In addition to perceived family support as the primary independent variable of interest, all measured sociodemographic characteristics and other factors having previously established or theoretically feasible associations with the dependent variables were included as covariates or potential confounders in the analyses, as listed within the table.

b

After accounting for the collective set of missing values associated with the variables included in the multivariable model, data from 292 participants were analyzed.

*p<0.05;

**p<0.01.