Background:  The CD4 count at which ART should be initiated is a central, unresolved issue in the care of HIV-1-infected patients. In the absence of randomized trials, we examined this question in prospective cohort studies.

Methods:  ARV-naïve patients from 15 cohort studies, who started ART after 1997 with CD4 <550 cells/mm³, AIDS-free and without a history of injection drug use, were eligible (ART Cohort Collaboration). We estimated distributions of lead-time (time from upper to lower CD4 threshold) and unseen events (events occurring before reaching the lower threshold) in the absence of treatment using data on 21,247 patients followed in seven cohort studies during the pre-ART era, and used these to impute completed datasets including lead-time and unseen events. We compared the effect of deferred with immediate initiation of ART on rates of AIDS and death, and death, in adjacent CD4 ranges of width 100 cells/mm³.

Results: The table shows estimated hazard ratios for the effect of deferring ART compared with immediate initiation, for different CD4 ranges. Deferring until CD4 count range 251 to 350 cells/mm³ was associated with higher rates of AIDS and death, compared with starting in the range 351 to 450 cells/mm³. The adverse effect of deferring ART increased with decreasing CD4 threshold. Deferring ART was associated with higher mortality rates, although effects on mortality were less marked than effects on AIDS and death.

Estimated hazard ratios for AIDS or death, comparing deferring ART to a lower CD4 range with starting in a higher CD4 range

Conclusions:  In the absence of evidence from a randomized controlled trial (the only design that can exclude the influence of unmeasured confounding) these findings should help guide physicians and patients balancing treatment benefits and toxicities in deciding when to initiate ART.