Survival in HIV-infected patients who have received zidovudine: Comparison of combination therapy with sequential monotherapy and continued zidovudine monotherapy

Background: Among patients who begin receiving zidovudine during intermediate-stage human immunodeficiency virus (HIV) infection, it is unclear whether changing to combination therapy (adding didanosine or zalcitabine) or sequential monotherapy (changing to didanosine or zalcitabine) significantly improves survival. Objective: To determine, among patients who began receiving zidovudine during intermediate-stage HIV infection, the differential effects of changing to combination therapy (zidovudine with didanosine or zalcitabine) or sequential monotherapy (with didanosine or zalcitabine) or continuing zidovudine monotherapy. Patients: 1077 HIV-seropositive men in the Multicenter AIDS (acquired immunodeficiency syndrome) Cohort Study who began receiving zidovudine before an AIDS-defining illness developed. Setting: University-affiliated clinics in Baltimore, Chicago, Los Angeles, and Pittsburgh. Design: Longitudinal cohort study. Treatment groups and important prognostic variables were modeled as timedependent covariates in Cox proportional hazards models. Measurements: Progression to AIDS and death. Results: Compared with patients receiving continued zidovudine monotherapy, patients receiving combination therapy had a 45% improvement in survival (relative risk, 0.55 [95% Cl, 0.41 to 0.74; P< 0.001]) and patients who changed to sequential monotherapy had a 32% improvement in survival (relative risk, 0.68 [Cl, 0.52 to 0.89; P = 0.005]). In the landmark analyses, the median prolongation of survival associated with changing therapy was, at best, 3 to 6 months. Survival curves converged at 3.5 years for the 50 cells/mm³, disease-stage landmark, at 4.4 years for the 100 cells/mm³ landmark, and at 4.9 years for the 150 cells/mm³ landmark. Mortality within these periods was 100%, regardless of treatment group or landmark. Conclusions: For patients who began receiving zidovudine during intermediate-stage disease, changing to either combination therapy or sequential monotherapy was associated with a statistically significant survival benefit compared with continuation of zidovudine monotherapy. The absolute increase in survival was modest, however, and long-term survival remained poor. Simultaneous timedependent adjustment for changes in therapy and in important prognostic variables is necessary to derive relatively unbiased estimates of treatment effects in observational studies of HIV infection.