Fig. 3

Modelled probability of viral suppression in ART-experienced patients in terms of time since treatment switch following detection of the Q151M mutation for patients with a baseline viral load of (a) 2000 copies/mL (≈ 10th centile), (b) 40,000 copies/mL (≈ 50th centile) or (c) 500,000 copies/mL (≈ 90th centile). Response is modelled according to presence or absence of a ritonavir-boosted protease inhibitor in the ART regimen at time zero, but patients were not censored at change to drug regimen in this analysis. The expected probability (solid line) and 95% credibility interval (dashed lines) from Bayesian fitting of sequential Weibull models for viral suppression and rebound are shown. DDI use was adjusted for in this analysis, but results are shown for patients not on DDI