Impact of ART and Sexual Behaviour Changes on HIV in Uganda
Impact of ART and Sexual Behaviour Changes on HIV in Uganda
Recent evidence from clinical trials have shown the positive impact that ART may have on reducing new infections (incidence). Our modelling work supports this evidence by also showing a reduction in incidence, unless partner turnover rates increase greatly (>50%). However, although incidence falls, results suggested that HIV prevalence will be higher with ART than without. Among the scenarios examined, HIV prevalence may decrease among men only in an extreme case which may not be achievable in practice. That is, ART enrolment would need to begin while still in HIV stage 1, the rate of ART enrolment would need to be greater than 0.50/year, sexual partner turnover would not change, and the probability of transmission while on ART would be just 0.5% per partnership from female to male (representing >95% reduction in transmission compared to no ART). Even in this case, 8 of the 11 sets of parameter values that we assessed suggested that HIV prevalence would be higher in the presence of ART than in its absence.
Despite the short duration in the initial HIV stage of infection, our model estimates that in the absence of ART about 36% of incidence comes from partnerships formed in which the HIV-infected partner was in the highly infectious stage 1. Initiating ART during this stage, if feasible, would therefore have the greatest impact on the epidemic.
Previous studies have suggested a range of potential outcomes to rolling out ART, from increasing to decreasing HIV prevalence. Although our work supports this range of potential outcomes to rolling out ART, in all but the most extreme scenarios our results consistently suggest that HIV prevalence will rise as a result of ART.
Results from this study come from a model that was parameterised and fit using data from Uganda. We expect that our results provide a good estimate of the impact of ART and corresponding sexual behaviour change on HIV epidemiologic trends in countries that are similar to Uganda in terms of HIV prevalence trends pre-ART and HIV modes of transmission (mainly heterosexual sex).
Results of any modelling study may be influenced by model assumptions. We have described our model assumptions in detail in the online supplementary appendix http://sti.bmj.com/content/90/5/423/suppl/DC1. One model assumption was that of no HIV/AIDS-related mortality after the initial period on ART. This is likely overly optimistic as non-perfect ART adherence speeds HIV progression and a higher mortality rate. If we had modelled a slightly higher mortality rate, even among those on ART, then estimated HIV prevalence after ART is introduced would be lower, as infected people would die at a higher rate. Also, we used a deterministic mathematical model. As such, stochastic events that may influence the plausible range of the impact of ART on HIV prevalence and incidence were not modelled.
In our best fitting model simulation, 36.2% of HIV incidence prior to ART introduction came from people in HIV stage 1. While this is similar to that estimated elsewhere, it is higher than that estimated by some. If more HIV incidence is attributable to higher HIV stages than we estimated, then ART enrolment in later stages would have a greater impact on incidence and subsequent prevalence.
Discussion
Epidemiologic Forecast
Recent evidence from clinical trials have shown the positive impact that ART may have on reducing new infections (incidence). Our modelling work supports this evidence by also showing a reduction in incidence, unless partner turnover rates increase greatly (>50%). However, although incidence falls, results suggested that HIV prevalence will be higher with ART than without. Among the scenarios examined, HIV prevalence may decrease among men only in an extreme case which may not be achievable in practice. That is, ART enrolment would need to begin while still in HIV stage 1, the rate of ART enrolment would need to be greater than 0.50/year, sexual partner turnover would not change, and the probability of transmission while on ART would be just 0.5% per partnership from female to male (representing >95% reduction in transmission compared to no ART). Even in this case, 8 of the 11 sets of parameter values that we assessed suggested that HIV prevalence would be higher in the presence of ART than in its absence.
Despite the short duration in the initial HIV stage of infection, our model estimates that in the absence of ART about 36% of incidence comes from partnerships formed in which the HIV-infected partner was in the highly infectious stage 1. Initiating ART during this stage, if feasible, would therefore have the greatest impact on the epidemic.
Previous studies have suggested a range of potential outcomes to rolling out ART, from increasing to decreasing HIV prevalence. Although our work supports this range of potential outcomes to rolling out ART, in all but the most extreme scenarios our results consistently suggest that HIV prevalence will rise as a result of ART.
Results from this study come from a model that was parameterised and fit using data from Uganda. We expect that our results provide a good estimate of the impact of ART and corresponding sexual behaviour change on HIV epidemiologic trends in countries that are similar to Uganda in terms of HIV prevalence trends pre-ART and HIV modes of transmission (mainly heterosexual sex).
Limitations
Results of any modelling study may be influenced by model assumptions. We have described our model assumptions in detail in the online supplementary appendix http://sti.bmj.com/content/90/5/423/suppl/DC1. One model assumption was that of no HIV/AIDS-related mortality after the initial period on ART. This is likely overly optimistic as non-perfect ART adherence speeds HIV progression and a higher mortality rate. If we had modelled a slightly higher mortality rate, even among those on ART, then estimated HIV prevalence after ART is introduced would be lower, as infected people would die at a higher rate. Also, we used a deterministic mathematical model. As such, stochastic events that may influence the plausible range of the impact of ART on HIV prevalence and incidence were not modelled.
In our best fitting model simulation, 36.2% of HIV incidence prior to ART introduction came from people in HIV stage 1. While this is similar to that estimated elsewhere, it is higher than that estimated by some. If more HIV incidence is attributable to higher HIV stages than we estimated, then ART enrolment in later stages would have a greater impact on incidence and subsequent prevalence.
Source...