Childhood malignancies in the EUROCARE study: the database and the methods of survival analysis.

This paper describes the database of children with cancer in the EUROCARE study and the methodology used to analyse and report survival. This is the first systematic evaluation of survival after childhood cancer on a large scale in Europe: approximately 45,000 cases were included, diagnosed between 1978 and 1992 (34,814 cases diagnosed in 1978--1989 and an additional set of 9495 cases diagnosed in 1990--1992) and followed-up until 1995. Data were provided from 34 population-based registries (four specialised for childhood cancer registrations and one specialised registry for childhood leukaemia) in 17 countries of Europe (where there was national coverage in 10 countries). Quality of the data was fairly good, given the general differences among the countries and their health systems, thereby allowing for comparisons between them. Among cases diagnosed in 1978--1989, overall 2.0% were lost to follow-up, 91.8% were microscopically diagnosed and 93.4% of alive cases had at least 5 years of observation. Survival proportions (observed survival) were calculated for each of the countries involved, by age group (0, 1--4, 5--9, 10--14 years), gender, different time periods and selected diagnostic groups. Age-standardised cumulative survival rates and European averages (weighted and pooled) were also computed. Cox regression models were used to evaluate geographical and temporal differences. The EUROCARE database represents a unique source of information on survival of childhood cancer patients in Europe, intercountry differences and time trends in survival.