Background:  Recent results from clinical trials (iPrEx, Partners PrEP, CDC Botswana) showed that oral pre-exposure prophylaxis (PrEP) based on tenofovir may be effective in preventing HIV while other trials (FEM-PrEP) were not able to replicate these positive findings. These discrepancies could be attributed to different trial populations and the lack of consistent PrEP use by the participants. The HIV acquisition risk among users with detectable drug levels in iPrEx was much lower than those without a detectable level. We aim to analyze the impact of adherence and the loss of PrEP protection between doses on the observed efficacy in randomized clinical trials with different dosing regimens (daily, every 2 days, every 3 days).

Methods:  We developed a stochastic mathematical model to simulate HIV transmission in a 2-arm clinical trial of oral PrEP. Several adherence scenarios are used, including skipping doses randomly. We used 2 functional expressions for the decrease in PrEP protection over time after a PrEP dose was taken:  a faster linear decrease, and a slower sigmoid decrease. We simulated different dose regimens and estimated the expected PrEP efficacy using 1000 simulations per scenario.

Results:  The dose regimen and the functional form of the decrease of the relative PrEP protection strongly affects the efficacy observed. Assuming an average individual PrEP efficacy of 60% and average adherence of 60% (both varied 40% to 80% uniformly) the expected effectiveness of daily, 2-day, and 3-day regimens with a linear loss of protection is 48% (90%CI 23 to 62), 34% (90%CI 13 to 55), and 23% (90%CI 4 to 46), respectively. The expected effectiveness with a sigmoid decrease is higher by 20% to 50% depending on the dosing regimen. The difference in the average adherence level between infected and uninfected participants at the end of the trial is negligible. However, the expected PrEP effectiveness among participants with adherence >70% is 60%, while among participants with adherence <50% is 40% assuming a daily regimen.

Conclusions:  Expected effectiveness of oral PrEP depends on how fast PrEP users lose protection between doses. This combined with poor adherence drives a significant difference in effectiveness for subgroups with high and low adherence and may explain the discrepancy in the results from different trials. The effect of adherence on effectiveness is more substantial with intermittent dosing regimen.