Extended Data Fig. 1: Timing of injections in the long-acting therapy group.

Timing of injections in 255 participants in the long-acting group in relation to the target date, until the time of withdrawal from the trial (3 participants), death (1 participant) or switch to standard treatment (for virological failure in 3 participants and adverse events in 2 participants). The target date for all injections was set at the time of randomisation (4, 8, 16, 24, 32, 40, 48, 56, 64, 72, 80, 88, and 96 weeks from randomisation). For one participant who interrupted all ART at day 7 (due to incarceration) and had the first injection day at day 125 (delay of 97 days), the target date was determined from the time of the first injection day. For 10 additional participants at one site who had injection delays of more than 14 days, from the time of each injection delay the target date for each subsequent injection was determined from the date of last injection plus 56 days (plus 28 days if the initial injection was delayed); with the target reset using the same principle at the time of each subsequent injection. There were 29 injections that were delayed > 14 days from the target date (occurring in 25 participants; 4 participants had two injections delayed > 14 days, none of which were consecutive injections). The interval from the target date to each delayed injection was 15 to 21 days for 20 injections; 22 to 28 days for 7 injections and > 28 days for 2 injections. Oral lead in with cabotegravir and rilpivirine was extended by 16 days in 1 participant to cover a period of travel before the initial injection; oral bridging was used to cover periods of travel in 2 participants when return to the site was not possible (one with cabotegravir and rilpivirine for 28 days and one with dolutegravir and rilpivirine for 19 days); and oral bridging was used to replace one injection to allow resolution of an injection site abscess in one participant (cabotegravir and rilpivirine for 56 days). Of the 25 participants with one or more injections delayed > 14 days from the target date, 22 had viral loads < 50 copies/ml at all measurement time points; three had a single measured viral load ≥ 50 copies/ml at one timepoint (66, 108 and 381 copies/ml) and all had VL < 50 copies/ml at week 96.