Table 1.
Multivariable Hazard Ratios and Regression Coefficients From the Best-Fitting Cox Proportional Hazards Model Used to Estimate Human Immunodeficiency Type 1 (HIV-1) Exposure Scores
| Characteristic | Hazard Ratio | 95% CI | P | Regression Coefficient |
|---|---|---|---|---|
| Couple | ||||
| Any unprotected sexa,b | 4.2 | (2.7–6.6) | <.001 | 1.4 |
| Seropositive partner | ||||
| HIV-1 RNA load (per 1 log10 increase)a | 2.7 | (2.1–3.5) | <.001 | 1.0 |
| Genital ulcer diseasea | 1.6 | (.9–2.8) | .08 | 0.5 |
| Seronegative partner | ||||
| Pregnancya | 1.8 | (.9–3.8) | .12 | 0.6 |
| HSV-2 seropositive | 2.1 | (1.2–3.8) | .01 | 0.8 |
| Male circumcision | 0.6 | (.3–1.1) | .10 | −0.5 |
| Age (10 y increase)c | ||||
| Among females | 0.4 | (.2–0.8) | −0.8 | |
| Among males | 0.8 | (.5–1.1) | −0.3 | |
| Female vs male sexc | ||||
| At 25 y of age | 0.9 | (.5–1.8) | −0.1 | |
| At 40 y of age | 0.5 | (.2–1.01) | −0.8 |
The best-fitting model was determined using backward variable selection with Akaike's information criterion as the stopping rule, which resulted in retention of some covariates with P > .05.
Abbreviations: CI, confidence interval; HSV-2, herpes simplex virus 2.
a Frequency of unprotected sex, plasma HIV-1 RNA levels, genital ulcer disease, and pregnancy were modeled as time-dependent variables. Baseline measurements from enrollment were used for all other covariates.
b Unprotected sex was indicated if either partner reported at least 1 occurrence of sexual intercourse without using a condom since the previous quarterly visit.
c Evidence of statistical interaction between the sex and age of the HIV-1–seronegative partner (P = .1) suggests that increasing age results in more substantial decrease in HIV-1 acquisition risk among females than males. The coefficients associated with age, sex, and the interaction were −0.2, 1.0, and −0.4, respectively.