Skip to main content

Table 2 Multivariate results showing the G-transformation hazard ratios for the individual, behavioral, and structural level covariates of HIV acquisition

From: HIV incidence declines in a rural South African population: a G-imputation approach for inference

 

Men

Women

 

HR

95% CI

P-value

HR

95% CI

P-value

Individual Variables

Age (vs. 15–19 years):

20–24

3.743

(3.131, 4.476)

<0.001

1.382

(1.250, 1.528)

<0.001

25–29

5.554

(4.714, 6.544)

<0.001

1.161

(1.033, 1.305)

0.012

30–34

6.175

(5.094, 7.486)

<0.001

0.990

(0.846, 1.158)

0.896

35–39

4.948

(3.704, 6.611)

<0.001

0.374

(0.296, 0.472)

<0.001

40+

2.523

(2.051, 3.104)

<0.001

0.242

(0.202, 0.289)

<0.001

Circumcised (vs. Uncircumcised)

0.729

(0.558, 0.953)

0.021

   

Behavioral Variables

Married (vs. Single)

0.619

(0.466, 0.823)

<0.001

0.396

(0.343, 0.459)

<0.001

Condom use: Sometimes (vs. Always)

0.973

(0.830, 1.140)

0.734

0.628

(0.574, 0.687)

<0.001

Cum. Outmigration (vs. Low): †

Moderate

0.883

(0.738, 1.056)

0.172

1.304

(1.186, 1.433)

<0.001

High

1.163

(0.976, 1.387)

0.091

1.197

(1.077, 1.330)

<0.001

Structural Variables

Household SES (vs. Lower tertile)

Middle tertile

0.847

(0.678, 1.060)

0.146

0.955

(0.856, 1.066)

0.412

Upper tertile

0.933

(0.776, 1.121)

0.459

0.999

(0.904, 1.104)

0.987

HIV Prevalence (vs. Low): ‡

Moderate

0.940

(0.566, 1.560)

0.811

1.361

(1.205, 1.536)

<0.001

High

1.325

(0.885, 1.983)

0.171

1.601

(1.413, 1.815)

<0.001

  1. Hazard ratio (HR), Standard error (SE), Confidence interval (CI). Hazard ratios were obtained with the G-transformation model for interval censored data
  2. †Low, moderate, and high defined as <5%, 5–20%, and >20% cumulative time spent outside of the study area for men and women
  3. ‡Opposite-sex HIV prevalence of the participant’s surrounding community. Low, moderate, and high prevalence defined as <10%, 10–20%, and >20% for men and <15%, 15–25%, >25% for women, respectively. Different categories used because of the large difference in HIV prevalence among men and women