Purpose: We performed a decision analysis to determine if a 57-year old female with a history of atrial fibrillation, but without prior systemic emboli, and ductal carcinoma in situ (DCIS) should receive lifelong anticoagulation and/or tamoxifen.
Methods: Using published data, we built a Markov model comparing four treatment strategies: combination therapy (warfarin plus tamoxifen); warfarin alone; tamoxifen alone; and neither warfarin nor tamoxifen. We considered risks of morbidity and mortality from major thromboembolism, major bleeding, and recurrent breast cancer that might occur over a lifetime (lifetime time horizon). To account for morbidity, we applied published quality of life measures.
Results: In the baseline analysis, when life expectancy alone was considered, treatment with warfarin alone yielded 21.7 years, compared with 21.5 years with neither warfarin nor tamoxifen, 21.3 years with combination therapy, and 19.6 years with tamoxifen alone. When quality of life was also considered, combination therapy yielded 19.7 quality-adjusted life years (QALYs), compared with 19.6 QALYs with warfarin alone, 19.5 QALYs with no medication, and 18.3 QALYs with tamoxifen alone. All parameters were varied over plausible ranges in sensitivity analyses. Combination therapy remained the preferred strategy over warfarin alone (other strategies were inferior) unlessthe annual probability of thromboembolism without tamoxifen was higher than 1.1% (baseline 1.0%), the annual probability of breast cancer recurrence was below 2.6% (baseline 3.0%), the efficacy of warfarin fell below 0.61 (baseline 0.65), the quality of life taking tamoxifen was below 0.99 (baseline 1.00), or the efficacy of tamoxifen was below 0.29 (baseline 0.32). Neither warfarin nor tamoxifen became preferred over all other strategies if quality of life while taking warfarin fell below 0.98 (baseline 0.99).
Conclusions: This analysis is consistent with current guidelines for breast cancer risk reduction with tamoxifen and reflects the complexity of the decision, particularly for this postmenopausal patient with an increased risk of thromboembolism secondary to atrial fibrillation. Tamoxifen alone was clearly the least preferred strategy; the choice among the other strategies was clearly a close call. Our analysis demonstrates how patient preferences and carefully individualized calculation of risks and benefits might inform the decision-making process and provide more patient-centered care.
Candidate for the Lee B. Lusted Student Prize Competition