Causes of the decline in AIDS deaths, United States, 1995-2002: prevention, treatment or both?

The decline in AIDS deaths in the USA between 1995 and 2002 has been attributed by Centers for Disease Control (CDC) to HIV treatments advances. The purpose of the present study is to assess whether this AIDS deaths decline was due entirely to treatment advances, to earlier prevention successes, or a combination of both. Secondarily, we quantitatively estimate the number of AIDS deaths averted (or delayed) by treatment advances over and above prevention effects. The study employed scenario analysis to address the research questions. To answer the primary research question, we examined whether three key predictions derived from the shape and peak of the HIV incidence curve in the USA (and the natural history of HIV disease in the era before highly active antiretroviral therapy [HAART]) about the shape and peak of the AIDS deaths curve were upheld (e.g., one prediction was that the peak of the AIDS deaths curve should occur 10-12 years after the peak of the HIV incidence curve). To the extent that these predictions are supported, there is evidence that earlier HIV prevention efforts impacted the number of AIDS deaths later in the epidemic. To answer the second research question, the observed annual AIDS deaths curve (1995-2002) was compared with three estimated AIDS deaths curves that may have occurred had HAART never became available. Three estimations were employed to reflect a range of assumptions about the lag between the flattening of HIV incidence in the USA and the flattening of AIDS deaths (i.e., 10, 11 or 12 years). For any one of the three-scenario analyses, the quantitative area between the 'observed' and 'estimated' AIDS deaths curves provide an estimate of the number of AIDS deaths averted by HIV/AIDS treatments. The three predictions from the HIV incidence curve (and the pre-HAART natural history of HIV disease) for determining the shape and peak of the AIDS deaths curve were supported thereby indicating the influence of past prevention efforts on recent AIDS deaths. However, the observed decline in AIDS deaths was more precipitous than predicted (indicating the influence of treatment advances). The scenario analyses indicated that between 33,480 and 41,784 AIDS deaths were averted (or delayed) between 1995 and 2002 as a function of treatment. That is, approximately, 206,037 AIDS deaths occurred between 1995 and 2002 (in the HAART era), but between 239,517 and 247,821 likely would have occurred without the advent of HAART. We conclude that both past prevention and current treatment services have lead to the AIDS deaths decline in the USA, and that HAART has averted between 33,480 and 41,784 AIDS deaths in the USA between 1995 and 2002 over and above the effects of early prevention efforts.