Document 1128
 DOCN  M94A1128
 TI    Should asymptomatic HIV positive patients take AZT: a decision analysis
       approach.
 DT    9412
 AU    Walmsley S; Naimark D; Toronto Hospital, University of Toronto, Ontario,
       Canada.
 SO    Int Conf AIDS. 1994 Aug 7-12;10(2):202 (abstract no. PB0823). Unique
       Identifier : AIDSLINE ICA10/94371447
 AB    OBJECTIVE: To develop a mathematical model to highlight the factors to
       consider in deciding when and whether asymptomatic HIV infected patients
       should begin AZT. METHODS: A decision analysis tree was designed with
       the objective to minimize toxicity of AZT treatment while maximizing
       survival and AIDS-free intervals. The baseline probabilities used in the
       model were derived from natural history studies and controlled clinical
       trials published in the literature. Utilities were derived by a
       consensus approach by HIV treating physicians. Analyses were conducted
       using the computer software SMLTREE 2.9. Sensitivity analysis around the
       major variables were conducted through the clinically plausible ranges.
       RESULTS: For a baseline CD4 count of 500 x 10(6)/L, the treat none
       strategy of the model is preferred. At this baseline, the expected
       utility is unaffected by a wide range of variation in most of the
       probability and utility estimates. For a baseline CD4 count of 200 x
       10(6)/L, the treat all strategy is preferred. Subtle variations in the
       majority of probability and utility estimates can affect the expected
       utility of the strategies. Using the baseline values, the treat-all
       option is preferred when the CD4 count drops below 250 x 10(6)/L.
       DISCUSSION AND CONCLUSION: This mathematic model can help clarify the
       competing risks and important variables governing the choice of
       management strategies for asymptomatic HIV infected patients considering
       AZT therapy.
 DE    Acquired Immunodeficiency Syndrome/MORTALITY  Computer Simulation
       Decision Support Techniques  Human  HIV Seropositivity/*DRUG THERAPY
       Models, Theoretical  Zidovudine/*THERAPEUTIC USE  MEETING ABSTRACT

       SOURCE: National Library of Medicine.  NOTICE: This material may be
       protected by Copyright Law (Title 17, U.S.Code).