Similar risks were also found when analyses were performed on the subset of patients followed up for

at least 1 year, and in those who had their last visit <2 years before the date of analysis (data not shown). Of note, in the latter set of patients, the RR associated with calendar year was even higher (RR=0.59, 95% CI 0.57–0.62; P<0.0001). We formally tested the interactions between calendar year and both mode of HIV transmission and ART status, using the whole study population. The inclusion of interactions between year and mode of transmission led to a significant improvement in the fit (log-likelihood P=0.00012). In detail, the effect of year in the various subgroups was as follows: RR=0.843 (95% CI 0.81–0.876) Smad inhibitor for heterosexual contact, RR=0.780 (95% CI 0.764–0.842) for other routes of infection, RR=0.89 (95% CI 0.87–0.92) for IDU, and RR=0.853 (95% CI 0.8179–0.886) for homosexual Target Selective Inhibitor Library mouse contact (P=0.01), suggesting that the immunological

benefit conferred by ART in IDU was significantly smaller than that observed for people who acquired HIV infection via sexual contact. The interaction between year and ART status also yielded a significant improvement in the log-likelihood (P=0.0007). The effect of year in the ART status strata was as follows: RR=0.84 (95% CI 0.81–0.86) for people on ART for ≥6 months; RH=0.89 (95% CI 0.86–0.92) for those on ART for <6 months; RH=0.89 (95% CI 0.85–0.94) for those on

an ART interruption; and RH=0.89 (95% CI 0.85–0.92) for ART-naïve patients. In the subset of patients previously on ART for ≥6 months (Table 2b), the decrease in the risk of having a CD4 count ≤200 cells/μL per more recent year appeared to be as rapid as in the main analysis. The RRs associated with the other covariates were consistent with those of oxyclozanide the main analysis. The evidence for an interaction between calendar year and mode of HIV transmission was confirmed in this subset of patients (P<0.0001). In a univariable Poisson regression, calendar year was again significantly associated with the probability of having a VL >50 copies/mL (this probability decreased from 66 to 40% from 1998 to 2008; RR=0.94, 95% CI 0.94–0.95; P<0.0001). Figure 1 (right panel) depicts annual trends overall and after stratifying for mode of transmission and ART status. When we stratified by mode of transmission, overall, the highest prevalence of poor virological prognosis was found in IDU (58%), followed by those infected via heterosexual contact (53%), those infected via homo/bisexual contact (51%) and those infected by other routes (46%). χ2 comparisons showed a significant difference among all groups (P<0.0001); however, this difference was no longer significant in the multivariable analysis.