A patient-level calibration framework for evaluating surveillance strategies: a case study of mammographic follow-up after early breast cancer

Objective: Currently all women who have completed their primary treatment for early breast cancer are invited to receive routine annual mammography. There is no randomized controlled trial evidence to support this schedule, and model-based analysis is required. This paper describes a novel data collection and model calibration process to analyze the cost-effectiveness of alternative follow-up schedules for early breast cancer survivors. Methods: A discrete event simulation model describes the progression of early breast cancer after the completion of primary treatment, representing impalpable and palpable recurrence and the detection of impalpable disease via follow-up mammography. Retrospective data from the South Australian Cancer Registry and clinical and administrative hospital databases were linked for 407 postmenopausal women diagnosed with moderate-prognosis early breast cancer from 2000 to 2008. These data formed the basis of a patient-level probabilistic calibration process. Results: For 50- to 69-year-old survivors, annual follow-up for 5 years, with visits every 2 years thereafter, appears to be cost-effective. For women aged 70 to 79 years at diagnosis, a surveillance schedule similar to general population screening (2 yearly) appears to be most cost-effective if high rates of adherence can be maintained. Conclusions: This study demonstrated the potential value of combining linked, retrospective data and decision analytic modeling to provide estimates of costs and health outcomes that are sufficiently robust to inform cancer clinical guidelines and individual patient decisions regarding appropriate follow-up schedules.