Skip to main content
. 2017 Jul 21;18:42. doi: 10.1186/s12889-017-4593-0

Table 3.

Correlates of moderate or greater self-perceived HIV risk

(1) All respondents (2) HIV-negative respondents
Demographics
 Age (years) 1.05 1.03
(0.97, 1.14) (0.94, 1.13)
 Education (years completed) 0.96 0.95
(0.82, 1.11) (0.80, 1.12)
 Married (ref: not married) 0.61 0.74
(0.29, 1.29) (0.30, 1.82)
 Employed (ref: unemployed) 0.77 0.79
(0.52, 1.15) (0.48, 1.28)
 Per capita monthly household income 1.00 1.00
(1.00, 1.00) (1.00, 1.00)
Sexual behaviour
 Condom used usually/always (ref: never/sometimes) 0.57** 0.65
(0.36, 0.89) (0.38, 1.11)
 Lifetime partners ≥3 (ref: 1 or 2) 2.39*** 2.68***
(1.53, 3.73) (1.54, 4.67)
 Recent partner ≥5 years older (ref: <5 years older) 1.73** 1.32
(1.09, 2.76) (0.75, 2.35)
 Had a sexually transmitted disease (ref: no) 0.89 1.00
(0.52, 1.53) (0.50, 2.02)
 Recent partner perceived to have other partners (ref: no) 1.48 1.62
(0.86, 2.53) (0.88, 3.01)
 Consumed alcohol in past 30 days (ref: no) 1.52 2.55**
(0.84,2.78) (1.21, 5.36)
Socio-psychological factors
 HIV knowledge score (0–4) 0.90 0.84
(0.68, 1.19) (0.60, 1.18)
 Heard male circumcision reduces male HIV risk (ref: no) 0.38*** 0.31***
(0.22, 0.64) (0.17, 0.59)
Stigmatising attitudes (ref: no stigma)
 Stigmatising response to 1 item 0.69 0.76
(0.37, 1.28) (0.37, 1.55)
 Stigmatising response to 2/3 items 0.53* 0.47*
(0.26, 1.09) (0.20, 1.12)
 Know someone living with HIV or who died of AIDS (ref: no) 2.27*** 2.42***
(1.36, 3.78) (1.28, 4.60)
 Ever tested for HIV (ref: no) 0.67* 0.23***
(0.43, 1.07) (0.13, 0.39)
 Know partner’s HIV status (ref: no) 0.82 0.69
(0.06, 11.10) (0.39, 1.20)
 Constant 2.27*** 1.60
(1.36, 3.78) (0.07, 35.30)
Observations 539 374

Estimates of logistic regression models examining correlates of HIV risk perceptions. Dependent variable = 1 for moderate/high risk and zero for no or low risk. Data come from the 2009 wave of CAPS, and individuals answering “don’t know” to the HIV risk perceptions query are excluded (about 12% of respondents). Estimates reflect odds ratios, with 95% CI in parentheses. Sample weights, correcting for survey design and non-response, used in estimating all models. Column 1 is the model for all black female respondents. Column 2 focuses only on those testing negative for HIV in 2009

*** p < 0.01

** p < 0.05

* p < 0.1