Using a population‐based registry to identify patterns of care in childhood cancer in Florida

Background. Unlike cancers occurring in adults, childhood cancers are distinguished by being primarily nonepithelial in origin and by their relative rarity. Even with the availability of registries such as the Surveillance, Epidemiology and End Results program of the National Cancer Institute or the Florida Cancer Data System for the State of Florida, there are potential biases that may affect the estimates of pediatric cancer incidence, studies related to elucidating patterns of care, and other epidemiologic studies. Methods. To evaluate the magnitude of these potential biases and elucidate the settings (pediatric cancer center versus non‐cancer center) in which childhood cancers are treated, the authors performed a retrospective study of childhood cancer in Florida. Results. Approximately 19% of childhood cancer cases (in patients 0–19 years of age) in Florida diagnosed from 1981 to 1986 were treated outside of identified pediatric cancer centers in the state. Children with Hodgkin disease and brain tumors represented 43% of these cases. Among those cases treated in pediatric cancer centers, 23% were treated by physicians other than pediatric oncologists. Children with brain tumors represented 28% of these cases. Of those treated by pediatric oncologists, 65% were eligible for a cooperative group protocol and 55% of these were enrolled. Conclusions. Population‐based registries are necessary for describing the full extent of childhood cancer, but they have limitations in demonstrating patterns of care. Consequently, generalization from the experience of pediatric cancer centers is questionable, and the opportunity to test and achieve advances in diagnosis and treatment may be subject to selection bias.