Background:  Due to the success of ART, mortality in people with HIV has declined to such an extent that it is relevant to ask if death rates in optimally treated people are higher than the general population. We compared mortality rates in well-controlled HIV-infected adults on cART in SMART and ESPRIT to mortality in the general population.

Methods:  Non-intravenous drug-using (IDU) patients were included. Follow up was included if the person currently had both suppressed viral load (≤400 copies/mL SMART, ≤500 copies/mL ESPRIT) and CD4 ≥350/µL, or had both in the past 6 months, and was aged 20 to 70 years. Standardized mortality ratios (SMR) were calculated by comparing deaths event data from the cART-alone control arms of ESPRIT and SMART to death rates from the Human Mortality Database stratified by country, age, and gender. 

Results:  Three thousand two hundred and fifty-four patients (662 [20%] female, median age at randomization 43 years [interquartile range (IQR) 37 to 50 years]) contributed 12,183 person-years of eligible follow up to the analysis (median follow up 3.0 years [IQR 1.9 to 5.5]). Sixty-two deaths occurred during follow up. Commonest cause of death was non-AIDS malignancy (12, 19%) followed by non-AIDs infections (6, 10%), substance abuse (5, 8%), and myocardial infarction/ischemic heart disease (MI/IHD) (4, 6%). Only 2 deaths (3%) were AIDS related. Cause of death was unknown, however, in 13 (21%). Overall, the SMR was 1.26 (95%CI 0.97 to 1.62). For patients with a CD4 count between 350 to 499/µL, mortality rate was increased compared to the background population (SMR 1.79, 95%CI 1.19 to 2.59). No evidence for increased mortality was seen in individuals with CD4 counts above 500/µL (Table). In a separate analysis we considered mortality if ever previously viral load undetectable with CD4 ≥350 cells/µL during the trial; SMR was 1.59 (95%CI 1.27 to 1.96).

Conclusions:  In HIV-infected patients on ART, with an undetectable viral load, who maintained or had recovery of CD4 counts to >500 cells/µL we identified no evidence for a raised risk of death compared to the general population. Caution is required in interpreting this due to a number of likely confounders; inclusion of people willing to take part in a clinical trial and who have achieved a CD4 count >500 cslls/µL may select for persons with better health-seeking and intervention-adherent behavior but, as people with HIV have different rates of smoking and other risk factors, confounding could result in bias in either direction. Finally, any long-term impact of HIV that is apparent only in old age cannot yet be evaluated.