Concordance of hospital-based cancer registry data with a clinicians' database for breast cancer.

Objective  Reliable information is essential to both clinical and policy decision making. We aimed to shed lights on the similarity and differences between a hospital-based cancer registry with a clinicians' database for breast cancer by comparing the registered data on the same year. Methods  We performed a head-to-head comparison of breast cancer cases extracted from the hospital-based cancer registry and the clinicians' database maintained by the Division of Breast Surgery at the National Cancer Center Hospital in 2004. Results  The hospital-based cancer registry reported 827 cases of newly diagnosed breast cancer patients in 2004, while the clinicians' database contained 366 surgically treated cases from 2004. Of these, 276 cases overlapped. Presence or absence of treatment modality was discordant in 15% for radiation therapy, 19% for chemotherapy, and 24% for hormone therapy between the two data sets. Furthermore, the recorded disease pathology was discordant in 13% for pathology and 28% for staging, with 22% for T-stage, 7% for N-stage, 7% for M-stage. Conclusions  Although information contained in hospital-based cancer registry and clinicians' database are generally accurate, some important differences were revealed as a result of varying interpretations of clinical information. Analyses of these data sets must be made with attention to details such as eligible patients, registered treatment, and timing of registration.