Background:  In HIV-infected persons, the proportions of death due to many AIDS-related conditions have decreased since the advent of HAART therapy.  There is concern that deaths from some non-AIDS-related causes are increasing as adverse outcomes from therapy or because people are living longer with HIV infection. We examined diagnoses present at death to determine trends in causes of death since the advent of HAART therapy. 

Methods:  Data were obtained from the Adult and Adolescent Spectrum of HIV Disease (ASD) study, an ongoing medical record-review observational cohort study in 11 U.S. cities that has followed over 54,000 HIV-infected persons since 1990.  We analyzed data from persons who died in 1992-2000, comparing diagnosis rates during 1992-1995 (pre-HAART) with those in 1996-2000 (HAART).  We restricted analysis to 6 sites that collected death certificate cause-of-death data.  We compared the proportions of death diagnoses for the 2 periods using stratified Cochran-Mantel-Haenszel statistics and controlling for age, sex, race/ethnicity, HIV transmission mode, and study site. 

Results:  From 1992 through 2000, we observed a total of 7188 deaths, 4870 (68%) during the pre-HAART period, and 2318 (32%) during the HAART period.  Compared with the pre-HAART period, proportions of deaths with the following conditions decreased for tuberculosis (TB) (OR 0.41; CI 0.3-0.6), non-TB mycobacterial infections (OR 0.7; CI 0.6-0.8), pneumocystis pneumonia (OR 0.7; CI 0.6-0.9), and toxoplasmosis (OR 0.7; CI 0.6-0.9).  Proportions of deaths increased for liver disease (odds ratio [OR] 1.7; 95% CI 1.4-2.1), non-Hodgkins lymphoma (OR 1.5; CI 1.2-2.0), cachexia/wasting (OR 1.3; CI 1.1-1.6), kidney disease (OR 1.3; CI 1.1-1.5), and sepsis (OR 1.2; CI 1.1-1.4).  Although not reaching statistical significance, the trend for ischemic heart disease (OR 1.9; CI 0.99-3.62) was suggestive of an increase in proportion of deaths. 

Conclusions:  This study demonstrates that since the advent of HAART there have been declines in proportions of deaths from some AIDS-related causes, likely due to both improved treatment and prophylaxis.  However, this study shows that there have been increasing proportions of some deaths from non-AIDS-related causes including liver and kidney disease, and possibly ischemic heart disease.  While these results may be partially explained simply by the declines in AIDS-related causes, data from this analysis should be used to guide research into the question of whether some of these increases may be due to adverse outcomes from HAART or due to the aging population of persons infected with HIV.