Background:  The optimal time to initiate HAART for asymptomatic HIV-infected individuals is uncertain. Emerging data about the benefits of earlier ART, such as improved response to therapy and preservation of immune function, suggest that initiating HAART earlier in the course of HIV disease may improve outcomes. Prior studies have been limited by insufficient size, follow-up, and methods to compare survival for patients who initiate HAART with those who defer treatment from the same CD4 cell count level.

Methods:  In a Canadian and US collaboration of observational cohorts, we examined all participants with a CD4 count >500 cells/mm³ between 1996 and 2006 who were free of clinical AIDS and ART-naïve. We compared the relative hazard (RH) of death for patients who initiated HAART at a CD4 count >500 with those who deferred using an inverse-probability weighted Cox proportional hazards analysis stratified by cohort and calendar-year, and adjusted for demographic and clinical characteristics. Patients who deferred HAART at a CD4 count >500 whose CD4 count fell to <350 without initiating HAART were censored at that time regardless of whether they subsequently started HAART.

Results:  The analysis currently includes 9174 patients who contribute 25,337 person-years of follow up; 2620 (29%) initiated HAART at a CD4 count >500 and the remaining 6553 (71%) deferred. The proportion of patients initiating HAART at a CD4 count >500 by year peaked in 1997 to 1998 (16%) and declined <10% by 2003; median (IQR) CD4 count at initiation was 674 (579 to 831), and 196 patients died. Among patients who deferred, median (IQR) CD4 count for the 602 who initiated HAART between 350 and 500 was 435 (397 to 472), and 271 died. Patients who deferred treatment at a CD4 count >500 had significantly higher risk of mortality than those who initiated HAART (RH 1.4, 95%CI 1.1 to 1.7; p = 0.008). Increasing age (RH 1.5 95%CI 1.4 to 1.7; p <0.001), but neither baseline HIV-1 RNA nor CD4 count (within range >500) were independent predictors of mortality.

Conclusions:  In this large North American cohort collaboration, we found 36% higher risk of death for patients who deferred rather than initiated HAART at a CD4 count >500. We adjusted for several prognostic variables, but the decision to start treatment may involve additional factors that could confound these results. Our findings support the initiation of HAART earlier in the course of HIV disease than currently recommended.