Background: In the last decades, prognosis of children with neuroblastoma has improved only slightly. Traditional estimates of survival reflect the survival experience of children diagnosed many years ago, and may thus not capture more recent progress in prognosis. We applied a new method of survival analysis, denoted period analysis, to provide more up-to-date estimates of long-term prognosis. We selected the cases diagnosed before the German neuroblastoma screening project in 1995, allowing to assess the method by comparing the 1994 projected survival estimates with the observations made today (2000).
Procedure: The data comes from the population based German Childhood Cancer Registry. We included all 1,353 children diagnosed with neuroblastoma below age 15 between 1980 and 1994. We derived 5-, 10-, and 15-year survival estimates using traditional analysis or period analysis as needed. Where possible we compared the period analysis estimates with the later obtained actual estimates. We showed trends in survival for the sample as a whole and for prognostic subgroups.
Results: Survival probabilities increased over time especially in the subgroups with poor prognosis. Short-term survival probabilities improved more than long-term survival probabilities. Evaluation of the period-analysis estimates showed them to provide accurate and timely projections of prognosis of newly diagnosed patients.
Conclusions: The results suggest major improvements in prognosis of children with neuroblastoma, even prior to the start of the German neuroblastoma screening project, especially in advanced disease. This could have been disclosed with the application of the period analysis method in 1995 even then with considerable accuracy. We recommend a more widespread application of this method especially in population-based cancer registries.
Copyright 2002 Wiley-Liss, Inc.