Methods to assess population effectiveness of therapies in human immunodeficiency virus incident and prevalent cohorts

Two methods are presented for measuring population effectiveness (i.e., reduction of disease in a population in which only some receive treatment) of antiretroviral therapy among human immunodeficiency virus (HIV)-infected men at risk for acquired immunodeficiency syndrome (AIDS) and followed between January 1, 1986, and June 30, 1999, in the Multicenter AIDS Cohort Study. Method I, requiring use of a seroincident cohort, estimates relative hazards of AIDS for persons at equal duration of infection. Method II, allowing use of a seroprevalent cohort, estimates relative hazards since the beginning of therapy eras for persons starting at equal levels of prognostic markers of disease stage (CD4 cell count and HIV type 1 RNA). The follow-up interval was divided into four calendar periods to characterize different eras of antiretroviral therapy. For method I, the relative hazards were 1.52 (95% confidence interval (CI): 0.93, 2.49), 0.91 (95% CI: 0.66, 1.26), and 0.30 (95% CI: 0.18, 0.51) for the eras of no therapy, dual nucleoside therapy, and potent combination antiretroviral therapy, respectively (monotherapy was the reference era). For method II, the corresponding relative hazards were 1.52 (95% CI: 1.10, 2.09), 1.03 (95% CI: 0.77, 1.38), and 0.31 (95% CI: 0.21, 0.45). These results extend the measurement of population effectiveness from incident to prevalent cohorts and demonstrate the ability of cohort studies to complement information provided by clinical trials.