Table 1.
Acquisition rates, and univariable and multivariable Cox regression model results for time to partnership acquisition (N = 587).
| Variable | Eventsa,b (N = 161) | Person-years at risk | Rate per 100 person-years) (95% CI) | Unadj. HRc | CI | Adjusted HRd | CI | Wald P value |
| ART group | ||||||||
| Pre-ART | 62 | 597.51 | 10.38 (7.96, 13.30) | 1.00 | 1.00 | |||
| ART-eligible | 99 | 947.15 | 10.45 (8.50, 12.73) | 1.07 | (0.76, 1.49) | 0.97 | (0.70, 1.35) | 0.87 |
| Sex | ||||||||
| Female | 120 | 1138.67 | 10.54 (8.74, 12.60) | 1.00 | 1.00 | |||
| Male | 41 | 405.98 | 10.10 (7.25, 13.70) | 0.96 | (0.64, 1.42) | 0.62 | (0.41, 0.94) | 0.025 |
| Age (years) | ||||||||
| 18–21 | 18 | 64.15 | 28.06 (16.63, 44.35) | 1.89 | (1.18, 3.02) | 1.96 | (1.25, 3.06) | <0.001 |
| 22–29 | 61 | 387.46 | 15.74 (12.04, 20.22) | 1.00 | 1.00 | |||
| 30–39 | 60 | 573.60 | 10.46 (7.98, 13.46) | 0.73 | (0.51, 1.04) | 0.71 | (0.50, 1.01) | |
| 40+ | 22 | 519.45 | 4.24 (2.65, 6.41) | 0.28 | (0.16, 0.50) | 0.21 | (0.12, 0.36) | |
| Partner status at enrolment | ||||||||
| Ongoing partnere | 100 | 1156.55 | 8.65 (7.04, 10.52) | 1.00 | 1.00 | |||
| No partner | 61 | 388.11 | 15.72 (12.02, 20.19) | 1.81 | (1.29, 2.52) | 2.50 | (1.84, 3.40) | <0.001 |
| Has previously disclosed HIV status to anyone | ||||||||
| No | 25 | 185.13 | 13.50 (8.74, 19.93) | 1.00 | 1.00 | |||
| Yes | 136 | 1359.53 | 10.00 (8.39, 11.83) | 0.74 | (0.45, 1.20) | 0.58 | (0.37, 0.89) | 0.014 |
| Ever used alcohol | ||||||||
| No | 63 | 820.91 | 7.67 (5.90, 9.82) | |||||
| Yes | 98 | 723.74 | 13.54 (10.99, 16.50) | 1.76 | (1.27, 2.45) | 1.70 | (1.18, 2.44) | 0.004 |
| No. of lifetime partnersf | ||||||||
| ≤3 | 75 | 907.40 | 8.27 (6.50, 10.36) | 1.00 | 1.00 | |||
| >3 | 81 | 600.64 | 13.49 (10.71, 16.76) | 1.63 | (1.18, 2.26) | 1.81 | (1.27, 2.58) | 0.001 |
ART, antiretroviral therapy; CI, confidence interval; HR, hazard ratio.
aIn a few instances, participants reported getting back together during follow-up with the person they had reported as their most recent but not ongoing partnership at enrolment. A few other participants reported a new partner but they never became sexually active with them. These were not considered new acquisitions in the analysis.
bTwenty-four partnerships were concurrent, that is, reported to have started while the participant was in at least one other ongoing partnership.
cNo other variables were significant in univariable models.
dThe final model did not violate the proportional hazards assumption, global test P = 0.09. Fitting a model with Gaussian individual frailties, the estimated variance was not found significantly different from zero (P value = 0.92), suggesting very little variation between individuals.
eThere were two groups of participants with ongoing partners at baseline, those who were recently sexually active and those who were abstaining from sex with their partners for various reasons. The number of new acquisitions, acquisition rates and 95% CIs for these two groups were: 90/1054.3 = 8.54 (6.86, 10.5) and 10/102.25 = 9.78 (4.69, 17.99) respectively and were not statistically different from each other.
fMissing values for 15 participants.