PURPOSE: The Surveillance, Epidemiology and End Results (SEER) contains information on breast cancer patient's demographic and tumor characteristics at the time of diagnosis and their vital status. Because it does contain information on the screening history or even mode of cancer detection and under-reports the use of adjuvant therapy, the SEER limits our ability to quantify the impact of cancer control programs. Through a collaboration with Cancer Intervention and Surveillance Network (CISNET), we have developed a simulation model that brings together data from numerous national registries to develop a virtual SEER database that captures information on the screening history, model of detection, tumor characteristic, treatment and survivorship on the individual patient level. We propose to present the underlying model that generates the virtual database and methods we used to validate it.

METHODS: We developed a Monte Carlo simulation model that reproduces the life history of women in the United States who were born since 1890 and outputs population level statistics from 1975 onward. The simulation records the following information for each breast cancer patient : her date of birth, her screening schedule, when and how she was detected, her tumor size and stage at detection stage, ER status at detection, what treatment she received, and her survival time from diagnosis and her cause of death. The model is informed with data from pre-screening era of SEER, the Patterns of Care Registry, and the Breast Cancer Surveillance Consortium.

RESULTS: Our virtual breast tumor registry closely reproduces population level trends observed in the SEER database that were not used in the model building process. The model reproduces the tumor size distribution as a function of calendar year for individual age groups. It reproduces the proportion of tumor stage (local, regional, distant) by calendar year and age-group. It demonstrates a good agreement between the model and data in terms of overall age-adjusted incidence and mortality.

CONCLUSION: We have generated and validated a virtual tumor registry of the US breast cancer population that merges existing national databases.