Background: Mortality and morbidity related to AIDS have decreased among HIV-infected patients taking highly active anti-retroviral therapy (HAART), but previous studies may have been confounded by other changes in treatment.

Objective: To assess the benefit of HAART in patients with advanced AIDS and anemia.

Design: Prospective, multicenter cohort study.

Setting: The Viral Activation Transfusion Study (VATS), with enrollment from August 1995 through July 1998 and follow-up through June 1999.

Patients: 528 HIV-infected patients with cytomegalovirus (CMV) seropositivity or disease who were receiving a first red blood cell transfusion for anemia.

Measurements: In a person-year analysis of follow-up before and after initiation of HAART, Poisson regression was used to calculate crude rate ratios and rate ratios adjusted for CD4 count, HIV RNA level, calendar period, time on study, sex, ethnicity, and injection drug use.

Results: At baseline, patients had a median CD4(+) lymphocyte count of 0.015 x 10(9) cell/L, median plasma HIV RNA level of 4.8 log(10) copies/mL, and median hemoglobin concentration of 73 g/L. Use of HAART increased from 1% of active patients in January 1996 to 79% of active patients in January 1999. The crude death rate was 0.24 event/person-year among patients taking HAART and 0.88 event/person-year among those not taking HAART (rate ratio, 0.26; adjusted rate ratio, 0.38; P < 0.001 for both comparisons). Rates of non-CMV disease were 0.15 event/ person-year after HAART and 0.45 event/person-year before HAART (crude rate ratio, 0.34 [ P < 0.001]; adjusted rate ratio, 0.66 [ P < 0.05]). Rates of CMV disease were 0.10 event/person-year after HAART and 0.25 before HAART (crude rate ratio, 0.42 [ P < 0.01]; adjusted rate ratio, 1.01 [ P > 0.2]). Results were similar in patients with baseline CD4(+) lymphocyte counts less than 0.010 x 10(9) cells/L.

Conclusions: The data support an independent reduction in mortality and opportunistic events attributable to HAART, even in patients with very advanced HIV disease. However, patients with CMV infection or disease may not have a reduction in new CMV events due to HAART.