Comparison of models and results from HIV dynamic transmission models used to evaluate the impact of male circumcision.

The first set of columns show the differences in model structure and the issues investigated in published analyses; the second set of columns show the results from each model making a standard set of assumptions.aModel simulations with universal or near universal uptake of male circumcision were compared to maximise comparability of outputs since high coverage levels would tend to exaggerate differences in model results. For White et al. [17], circumcision is scaled up from 25% to 75%. Simulated coverage reaches scale within 5 to 10 y for all models. The impact on HIV infections averted with more realistic coverage levels is shown in Figure 1. Where possible, the following standard sets of assumptions are used: 60% reduction in female-to-male transmission, 0% reduction in male-to-female transmission associated with HIV, and no risk compensation.bBasic reproductive number approximately equal to one, indicating that the epidemic would be on the verge of terminal decline and incidence declines gradually.cThis estimate is based on a simplified model for low risk populations. In the revised and expanded model [16], the estimate for the number of operations per infection averted over the first 10 y of the intervention in a population representative of Rakai, Uganda is 11.eq, equilibrium.