Background: Mortality associated with HIV disease has decreased markedly since 1996 and changes in causes of mortality have been documented in some studies. We evaluated trends in mortality, location of death and causes of death in a well-defined cohort of patients (HAVACS).

Methods: All patients seen at the Atlanta VA since 1981 have been followed in a prospective cohort study. Information on deaths is determined in a hierarchical method following deaths occurring in the hospital (in-house), in the local region and reported automatically through the VA HIV Registry (ICR), through a national VA death benefits program (HINQ) and through the Social Security Death Index/ National Death Index (SSDI). Cause of death was evaluated from medical records, laboratory results or through death certificates.

Results:  From 1981 until September 23, 2003, we have seen 2502 HIV-infected patients at the Atlanta VA; 1122 (44.8%) have died. Overall 352 deaths occurred in-house, representing only 31.3 % of all deaths. Other sources of information included ICR (504), HINQ (224), and SSDI (41); VA sources of mortality data accounted for 96.4% of all deaths. Mortality rates decreased markedly from 25.9 per 100 in 1994 to 6.5 per 100 in 1996 and have been stable since that time. Cause of death has also changed significantly. AIDS/ opportunistic infections accounted for 50.0% of deaths from 1982 through 1996, but account for only 36.0% of deaths since that time (p = 0.00002). Despite a prevalence of HCV co-infection ranging from 33 to 35% in our population, end-stage liver disease (ESLD) accounted for only 55 deaths (4.9%) although there were more deaths after 1996 (28 of 412, 6.8%) than before 1996 (27 of 710, 3.8%, p = 0.034). Myocardial infarction or sudden death was diagnosed in only 18 of 1122 deaths (1.6%), but was more common after 1996 (13 of 412 [3.2%]) than before 1996 (5 of 710 [0.7%]), (p = 0.004). The incidence of both ESLD and myocardial infarction (per 100 patients) was the same before and after 1996.

Conclusions:  In this well-defined cohort, we demonstrated a marked decrease in mortality rate since 1996. A minority of deaths occurred in the hospital, but VA sources of mortality data were excellent and could document almost all deaths.  A minority of deaths has been due to AIDS/opportunistic infections since 1997. Continued careful analysis of specific causes of death in our population will be important to address the importance of comorbid conditions in recent mortality.

Keywords: mortality; opportunistic infections; end stage liver disease