Comparison of standard study design and serodiscordant couples study design for estimating the per-sex act risk of HIV infection among HSV2 infected individuals compared to HSV2 uninfected individuals
| Null Median HR (Lowest HR, Highest HR) | Different transmissibility Median HR (Lowest HR, Highest HR) | Coinfection increases hiv infectiousness Median HR (Lowest HR, Highest HR) | Susceptibility Median HR (Lowest HR, Highest HR) | |
| Expected | 1 | 1 | 1 | 10 |
| Standard Design | 1.32 (0.84, 2.26) | 1.58 (0.64, 1.44) | 1.26 (1.01, 1.56) | 4.38 (3.28, 5.65) |
| Serodiscordant Couples | 1.03 (0.84, 1.44) | 1.07 (0.70, 2.11) | 1.03 (0.72, 1.25) | 10.33 (9.65, 12.77) |
20 simulations were performed for each bias scenario. Simulations were started in 1930. Cohort studies were conducted from 1992-95, which is the period during which the data used to parameterise the model were collected in rural Tanzania. The data from each run was analysed using Cox proportional hazards regression in two ways, both of which enrolled sexually active index subjects who were HIV negative and who could be HSV2 positive or negative:
1) Standard design – data on exposure to infection was not incorporated into the analysis.
2) Serodiscordant couples design – only those individuals who had an HIV infected partner were included and only for the length of time they were in the partnership.
All analyses adjusted for age in 5-year intervals, gender, and the baseline number of lifetime partners. Results presented are the median, lowest, and highest point estimates for the HR calculated from these runs. HIV and HSV2 per-sex act transmission probabilities were both equal to 0.02 unless otherwise stated.
Null: No interaction between HSV2 and HIV.
Different Transmissibility: No interaction between HSV2 and HIV. HSV2 per-sex act transmission probability = 0.03.
Coinfection Increases HIV Infectiousness: Coinfected individuals have twice the HIV per-sex act transmission probability than individuals infected with HIV alone (ie, 0.04 vs 0.02 respectively).
Susceptibility: The HIV per-sex act transmission probability to an HSV2 infected individual is 10 times higher than to an HSV2 uninfected individual (ie, 0.2 vs 0.02 respectively).
Note: Susceptibility is not affected in the first three scenarios. Also, parameter values have been changed from disease specific values to permit each bias to be presented separately. Concurrent partnerships are not permitted. Full sensitivity analysis results which relax these restrictions are not shown.