Cost effectiveness analysis of early zidovudine treatment of HIV infected patients.

OBJECTIVE--To compare cost effectiveness of early and later treatment with zidovudine for patients infected with HIV. DESIGN--Markov chain analysis of cost effectiveness based on results of use of health care and efficacy from a trial of zidovudine treatment. SETTING--Seven Veterans Affairs medical centres in the United States. SUBJECTS--338 patients with symptomatic HIV infection and a lymphocyte count of 200 x 10(6) to 500 x 10(6) CD4 cells/l. INTERVENTIONS--Zidovudine 1500 mg/day started either at recruitment to the trial or when CD4 cell count fell below 200 x 10(6)/l. MAIN OUTCOME MEASURES--Health care costs and rates of disease progression between six clinical states of HIV infection. RESULTS--Patients given early treatment with zidovudine remained without AIDS for an extra two months at a cost of $10,750 for each extra month without AIDS (at 1991 costs). Cost effectiveness ratio was most sensitive to the cost of zidovudine and to the quality of life of patients receiving early treatment. At treatment of 500 mg/day the cost effectiveness ratio for early treatment was $5432 for each extra month without AIDS. Patients given early treatment experienced more side effects, and if their quality of life was devalued by 8% compared with patients treated later the two treatments were equivalent in terms of quality adjusted months of life without AIDS. CONCLUSIONS--Early treatment with zidovudine is expensive and is very sensitive to the cost of zidovudine and to potential reductions in quality of life of patients who experience side effects. Doctors should reconsider early treatment with zidovudine for patients who experience side effects that substantially compromise their quality of life.

Duke Scholars

Author Eugene Zaverio Oddone Medicine, General Internal Medicine

Author John David Hamilton Medicine, Infectious Diseases

Author David Bruce Matchar Medicine, General Internal Medicine

Author Gregory P. Samsa Biostatistics & Bioinformatics, Division of Biostatistics