Figure 2.

Risk of virologic failure or death by arms over 2 years (conventional ITT risks, dropout-weighted ITT risks, per-protocol risks that accounted for informative dropout and protocol deviations in the A5095 trial and in the target population). ITT: intention-to-treat; target: target population.

Solid line represents the three-drug arm, and dotted line represents the four-drug arm. Upper panels represent risks in the trial with the left panel depicting the conventional ITT results under the independent censoring assumption, middle panel depicting the ITT results that accounted for informative dropout, and the right panel depicting per-protocol results that accounted for informative dropout and protocol deviation. Lower panels represent risks in the target US recently HIV-diagnosed population with the left panel depicting the conventional ITT results that only accounted for sampling bias, middle panel depicting the ITT results that accounted for informative dropout and sampling bias, and the right panel depicting per-protocol results that accounted for informative dropout, protocol deviation and sampling bias.