Thus, the effect of previous virological failure on current CD4 cell count persisted beyond 1 year. The effects of virological failure during the past year on CD4 cell counts (Table 3) were only slightly attenuated by controlling additionally for cumulative years of virological failure. Model 2 of Table 3 shows estimated effects of treatment interruption, before controlling for virological failure. Treatment interruption was associated with lower subsequent CD4 cell

counts, with the greatest adverse effects occurring 0–44 days after a treatment interruption. For the remaining three time periods, the size of the adverse effects were modest. In Table 3, model 3, the effects of virological failure and treatment HDAC inhibitor interruption were adjusted for each other. While the effects of virological failure were slightly attenuated, the effects of treatment interruptions were markedly attenuated, with ratios of geometric means close to 1 for all but the period 0–44 days before the current time. We further investigated whether the effects of virological failure differed between the 5113 participants who maintained treatment from 6 months since the start of cART to the end of follow-up, and those 1956 participants who experienced at least one check details treatment interruption. Of these 1956 participants, there were 970 with no

measured virological failure from 6 months after the start of cART, among whom the median total time a participant was off three or more antiretrovirals was 7 months [interquartile range (IQR) 2–16 months], the median number of HIV-1 RNA measurements until the end of follow-up was 16 (IQR 10–22) and the median baseline HIV-1 RNA was 82 768 copies/mL (IQR 19 352–256 000 copies/mL). In comparison,

among the 986 participants who experienced at least one treatment interruption and had a measured virological failure Carbohydrate 6 months after the start of cART, the median total time off three or more antiretrovirals was 13 months (IQR 5–27 months), the median number of HIV-1 RNA measurements until the end of follow-up was 24 (IQR 16–33) and the median baseline HIV-1 RNA was 73 300 copies/mL (IQR 17 614–272 000 copies/mL). The estimated effects of virological failure in those who had at least one treatment interruption were mainly slightly larger (smaller ratios of geometric means) than in those who maintained treatment. Each set of results was similar to those reported in Table 3 (available on request). Using data from a large, well-characterized cohort study, we have shown that, among patients who maintained viral load suppression, there were continuing increases in CD4 cell counts between 4 and 8 years after starting cART, regardless of CD4 cell count at initiation of cART. Nonetheless, differences in post-cART CD4 cell counts between baseline CD4 groups persist up to 8 years after initiation.