Background:  Adherence to antiretroviral drugs is associated with virologic response to HAART. It remains to be determined whether and at what extent adherence is a predictor of clinical progression, when other known clinical and social predictors are also considered.

Methods:  The APROCO cohort enrolled 1281 patients at the initiation of a protease inhibitor (PI)-containing regimen in 1997-1999. Patients are followed every 4 months. Predictors of clinical progression, defined as occurrence of AIDS or death, were studied using multivariate Cox models. Adherence and depression (CES-D scale) were measured using self-administered questionnaires which also collected information about social factors.

Results:  Of 1028 patients with more than 4 months of follow-up and 1 measure of initial adherence (at month 4), 124 (12%) reported initial low adherence. During a median follow-up of 54 months, 50 (5%) patients died and 52 (5%) had an AIDS-defining event. In univariate analysis, initial low adherence was associated with a higher risk of clinical progression (hazard ratio = 1.7 for patients with low adherence compared to others, p = 0.04). The multivariate Cox model I included baseline clinical-biological factors and initial adherence which was associated with progression with an adjusted hazard ratio (aHR) of 1.7 (p = 0.05). Other predictors of progression in model I were older age, CDC stage C and co-infection with hepatitis C virus. When early virologic failure defined as plasma HIV RNA >10 000 copies/mL at month 4 was added to model I, it was strongly associated with progression (aHR = 4.5, p <0.0001 for virologic failers vs others), while initial adherence was not (aHR = 1.2, p = 0.62). When social factors measured at baseline were added to model I, poor housing (aHR = 1.8, p = 0.009) and lack of steady partner (aHR = 2.0, p = 0.001) were associated with progression, whereas the association between initial low adherence and progression was lowered (aHR = 1.5, p = 0.11). When baseline depression was included in model I, both low adherence at month 4 (aHR = 2.1, p = 0.008) and depression (aHR = 1.7, p = 0.02) were associated with progression.

Conclusions:  Patients with low adherence, depression, poor social status or lack of a steady partner in the early course of HAART have a higher risk of progression to AIDS or death during long-term follow-up. These results clearly indicate the importance of introducing a screening for depression to timely treat patients who may need it.