Evidence-based surveillance strategies for melanoma survivors do not exist. Given the increasing incidence of melanoma and the rapid advances in health care technology, the costs of caring for such persons are rising. More than 90% of patients with melanoma are diagnosed before the development of distant metastasis and are potentially curable. However, up to 50% of all melanoma patients may experience tumor recurrences. The early detection of recurrent melanoma, particularly locoregional disease, at a time when it is amenable to surgical resection is important to improve patient outcome. Decision-analytic models which can assimilate data from a number of sources provide an innovative approach to examining outcomes in a setting where clinical trials are difficult to perform. We propose to use a decision-analytic Markov model created to simulate the stage-specific natural history of contemporary patients with melanoma. By applying techniques of Bayesian meta-analysis we will assimilate patient-level data from the literature to determine the sensitivity of various diagnostic imaging modalities. Various surveillance strategies, including those proposed in current protocols, (e.g., clinical evaluation, nodal ultrasonography, computed and positron emission tomography) and follow-up intervals (3 months to 2 years) will be evaluated and compared in patients with stages I, II, and III melanoma. Specific effectiveness outcomes will include: the number and type of recurrences detected, number of lives saved, and quality-adjusted life years. In addition, the total lifetime costs per surveillance strategy, incremental cost-effectiveness, and net benefit will be defined for each stage-specific strategy. To evaluate model and parameter uncertainty, a probabilistic sensitivity analysis will be performed. SEER data and costs from Medicare will be used to calibrate the natural history model and to examine the outcomes of various surveillance strategies in the population. Collectively, the outcomes from this proposal will provide evidence-based guidelines for surveillance, which in turn will promote the cost-effective use of health care resources. ? ? ?
