Ears. PrEP adherence. Adherence is key in PrEP use as illustrated by all recent PrEP studies [2,3,4,5]. Since it is unknown what level of adherence would be expected in Macha, we examined a high population-level adherence scenario and ranged PrEP effectiveness from 50 ?0 , derived from the highly adherent in recent PrEP trials [2,3,4], and a moderate population-level PrEP adherence scenario, where effectiveness ranged from 20 ?0 . Drug resistance. Rates of drug resistance due to PrEP are currently unknown. Drug resistance may emerge in individuals who become infected with HIV despite the use of PrEP. It is unknown how rapidly resistance will emerge after PrEP failure. We therefore evaluated a scenario with low resistance development, where resistance develops in 10 of breakthrough infections (infections despite the use of PrEP). We also evaluated a moderate resistance and high resistance scenario, where resistance emerges in 50 and 100 of breakthrough infections respectively. TheSensitivity AnalysisWe performed one-way deterministic sensitivity analysis of costeffectiveness where our baseline model for comparison was the prioritized PrEP model with moderate PrEP adherence. Eight key input variables, HIV prevalence, PrEP efficacy, proportion of people in highest two sexual activity groups on PrEP, number of HIV tests per year for those on PrEP, cost of antiretroviral drugs, total costs depending on the exchange rate, cost and QALY discounting were MedChemExpress BIBS39 considered to identify the sensitivity of our model. We also determined the amount of additional money that could be spent on infrastructure and programmatic costs of implementing prioritized PrEP and have the intervention still be (very) cost-effective.Ethics StatementWritten informed consent was obtained from the study participants. Ethical approval was granted by the University of Zambia Biomedical Research Ethical Committee in 2008 before data collection began.Results Baseline Scenario: Start of Treatment at CD4,350 Cells/ mmThe impact of treatment alone under the current guidelines of treatment at CD4,350 cells/mm3 reduces incidence, showing an 18 decline in new infections over 10 years. The prevalence remained stable at 7.7 after 10 years, as treatment 1081537 dramatically reduces mortality and patients therefore remain alive.Cost-Effectiveness of PrEP, ZambiaCost-Effectiveness of PrEP, ZambiaFigure 1. Prioritizing highest sexual risk groups versus a non-prioritized PrEP strategy, incidence and prevalence. doi:10.1371/journal.pone.0059549.get SC 1 gPrioritized Versus Non-Prioritized PrEPCompared to our baseline scenario of starting treatment at CD4,350 cells/mm3, prioritizing PrEP will result in 3200 infections averted over 10 years (31 reduction; interquartile range (IQR) 23 ?9 ), whereas a non-prioritized PrEP strategy will 
result in just 2333 infections averted (23 reduction; IQR: 16?0 ) (Figure 1A, 1E). The prevalence in the prioritized approach is lower after 10 years, at 5.7 (IQR: 5.2 ?.2 ), compared to a prevalence of 6.4 (IQR: 6.0 ?.7 ) in the nonprioritized strategy (Figure 1B, 1F).Impact of AdherenceAs expected, high PrEP adherence had a strong impact on the HIV epidemic as compared to moderate PrEP adherence in boththe prioritized and non-prioritized strategies. The impact, however, was stronger than expected. In the non-prioritized strategy, compared to baseline, an estimated 4333 infections (42 reduction; IQR: 35 ?0 ) were averted with high adherence to PrEP (Figure 1C), 2000 more t.Ears. PrEP adherence. Adherence is key in PrEP use as illustrated by all recent PrEP studies [2,3,4,5]. Since it is unknown what level of adherence would be expected in Macha, we examined a high population-level adherence scenario and ranged PrEP effectiveness from 50 ?0 , derived from the highly adherent in recent PrEP trials [2,3,4], and a moderate population-level PrEP adherence scenario, where effectiveness ranged from 20 ?0 . Drug resistance. Rates of drug resistance due to PrEP are currently unknown. Drug resistance may emerge in individuals who become infected with HIV despite the use of PrEP. It is unknown how rapidly resistance will emerge after PrEP failure. We therefore evaluated a scenario with low resistance development, where resistance develops in 10 of breakthrough infections (infections despite the use of PrEP). We also evaluated a moderate resistance and high resistance scenario, where resistance emerges in 50 and 100 of breakthrough infections respectively. TheSensitivity AnalysisWe performed one-way deterministic sensitivity analysis of costeffectiveness where our baseline model for comparison was the prioritized PrEP model with moderate PrEP adherence. Eight key input variables, HIV prevalence, PrEP efficacy, proportion of people in highest two sexual 
activity groups on PrEP, number of HIV tests per year for those on PrEP, cost of antiretroviral drugs, total costs depending on the exchange rate, cost and QALY discounting were considered to identify the sensitivity of our model. We also determined the amount of additional money that could be spent on infrastructure and programmatic costs of implementing prioritized PrEP and have the intervention still be (very) cost-effective.Ethics StatementWritten informed consent was obtained from the study participants. Ethical approval was granted by the University of Zambia Biomedical Research Ethical Committee in 2008 before data collection began.Results Baseline Scenario: Start of Treatment at CD4,350 Cells/ mmThe impact of treatment alone under the current guidelines of treatment at CD4,350 cells/mm3 reduces incidence, showing an 18 decline in new infections over 10 years. The prevalence remained stable at 7.7 after 10 years, as treatment 1081537 dramatically reduces mortality and patients therefore remain alive.Cost-Effectiveness of PrEP, ZambiaCost-Effectiveness of PrEP, ZambiaFigure 1. Prioritizing highest sexual risk groups versus a non-prioritized PrEP strategy, incidence and prevalence. doi:10.1371/journal.pone.0059549.gPrioritized Versus Non-Prioritized PrEPCompared to our baseline scenario of starting treatment at CD4,350 cells/mm3, prioritizing PrEP will result in 3200 infections averted over 10 years (31 reduction; interquartile range (IQR) 23 ?9 ), whereas a non-prioritized PrEP strategy will result in just 2333 infections averted (23 reduction; IQR: 16?0 ) (Figure 1A, 1E). The prevalence in the prioritized approach is lower after 10 years, at 5.7 (IQR: 5.2 ?.2 ), compared to a prevalence of 6.4 (IQR: 6.0 ?.7 ) in the nonprioritized strategy (Figure 1B, 1F).Impact of AdherenceAs expected, high PrEP adherence had a strong impact on the HIV epidemic as compared to moderate PrEP adherence in boththe prioritized and non-prioritized strategies. The impact, however, was stronger than expected. In the non-prioritized strategy, compared to baseline, an estimated 4333 infections (42 reduction; IQR: 35 ?0 ) were averted with high adherence to PrEP (Figure 1C), 2000 more t.