OBJECTIVE: To determine the long term benefits and costs of primary (tamoxifen) and secondary (mammography) breast cancer preventive strategies in healthy women who are at high risk of developing breast cancer.
METHODS: A Markov process with time and state dependent transition probabilities was developed. Three hypothetical cohorts of high risk women were initiated at age 40 and were simulated over their lifetime. Cohort one consumed tamoxifen and underwent routine mammography screening. Cohort two consumed tamoxifen but underwent mammography screening at rates observed in clinical practice (sub-optimal compliance) and Cohort three, which served as the control, did not consume tamoxifen and underwent mammography screening at rates observed in clinical practice (sub-optimal). Chemopreventive tamoxifen effect was modeled based on the five-year Breast Cancer Prevention Trial results while mammography efficacy data was based on data from published analysis which investigated effect of mammography screening on mortality. All cause and disease specific mortality rates were based on data calculated from life tables and other published sources. All costs and efficacy were determined using 3% discount rate.
RESULTS: Tamoxifen coupled with routine mammography screening resulted in an incremental life expectancy gain of 0.122 years at an incremental cost of $5969.7 resulting in an ICER of $48,931.8/life year gained compared to control group. Tamoxifen coupled with mammography screening at rates observed in clinical practice resulted in an ICER of $64,695.2/life year gained. Sensitivity analysis involving varying intervention effectiveness, adverse events resulting from tamoxifen, intervention compliance, and mortality rates associated with different health states, revealed the robustness of results over a wide range of assumptions.
CONCLUSIONS: Chemopreventive tamoxifen coupled with routine mammography screening in women who are healthy but at high risk for developing breast cancer was found to be a costeffective strategy. The results were robust over a wide range of assumptions.
