TAMOXIFEN DECISION MAKING: WHICH UTILITIES TO CHOOSE?
Joy Melnikow, MD, MPH, University of California, Davis, Sacramento, CA, Christina Kuenneth, MPH, University of California, Davis, Sacramento, CA, L. Jay Helms, PhD, University of California, Davis, Davis, CA, Miriam Kuppermann, PhD, University of California, San Francisco, San Francisco, CA, and Stephen Birch, DPhil, McMaster University, Hamilton, ON, Canada.
Purpose: Quality of life is a key element in the evaluation of the cost effectiveness of chemoprevention. Tamoxifen is the prototype chemo-preventive agent for breast cancer. We developed a Markov model to evaluate the cost-effectiveness of tamoxifen chemoprevention. Methods: Baseline population incidence of relevant outcomes was estimated from review of the medical literature. Relative risks were obtained from published randomized controlled trials. Costs were derived from Medicare reimbursements for clinical pathways defined for each outcome. Standard gamble utilities were measured in 217 women aged 50-90 whose five year estimated breast cancer risk was >1.67%, making them potentially eligible for tamoxifen chemoprevention. Following a 15 minute educational session about the potential benefits and harms of tamoxifen, utilities were obtained from all women for current health and their global assessment of taking tamoxifen for chemoprevention. Rotating questions assessed utilities for breast cancer, endometrial cancer, pulmonary embolism, deep venous thrombosis, hip fracture, and cataracts, so that 30-35 women provided a value for each state. Time trade-off utilities were obtained from the medical literature for hot flashes, a transient health state worsened by tamoxifen. The duration of disutility for each state was estimated from clinical experience and SEER data. Analyses were conducted with the global utility model and the outcome-specific model, with and without disutility for hot flashes. Sensitivity analyses varied breast cancer risk, age, hysterectomy status, and relative risks. Results: Employing the global utility model, tamoxifen was dominated at all breast cancer risk levels. The base case outcome-specific utilities model cost per QALY for women with a uterus fell below $100,000 per life year saved for women whose 5-year breast cancer risk exceeded 2%. For women without a uterus, the cost per QALY at a 5-year breast cancer risk of 1.67% was $72,530. Inclusion of disutility for hot flashes in the model increased the cost per QALY for this same group to $300,030 and tamoxifen was dominated for women with a uterus until their breast cancer risk exceeded 2.3%. Conclusions: Conventional methods of specifying utilities for cost-effectiveness analysis often neglect the effect of transient unpleasant states on quality of life. Consideration of utilities for these states may have profound effects on cost-effectiveness ratios.