Updating long-term childhood cancer survival trend with period and mixed analysis: Good news from population-based estimates in Italy Luisa Zuccolo a,b, *, Elisa Dama a , Milena Maria Maule a , Guido Pastore a,c , Franco Merletti a , Corrado Magnani a,d a Childhood Cancer Registry of Piedmont, Cancer Epidemiology Unit, CPO Piemonte, CeRMS, S. Giovanni Hospital and University of Turin, Italy b Department of Social Medicine, University of Bristol, Bristol, UK c Division of Pediatrics, Department of Medical Sciences, University of Eastern Piedmont, Novara, Italy d Unit of Medical Statistics and Epidemiology, CPO Piemonte and Department of Medical Sciences of the University of Eastern Piedmont, Novara, Italy ARTICLE INFO Article history: Received 15 July 2005 Received in revised form 13 August 2005 Accepted 28 August 2005 Available online 21 April 2006 Keywords: Childhood neoplasms Leukaemia Central nervous system neoplasms Survival analysis Cancer registries Population-based studies Statistical methods ABSTRACT An empirical evaluation of long-term period survival analysis was performed using data from the Childhood Cancer Registry of Piedmont, Italy. The aim was to update survival time trends and provide 25-year projections for children currently diagnosed with cancer. The observed survival experiences up to 15 years after diagnosis of five quinquennial cohorts (cohort analysis) were compared to the corresponding estimates obtained by period anal- ysis. The two methods generally produced very similar findings, although period analysis estimates were slightly lower than those obtained from cohort analysis. We then used mixed analysis to assess time trends in long-term survival. This showed that the probabil- ity of surviving 25 years after a cancer in childhood has more than doubled compared to cohort analysis estimates from patients diagnosed more than 25 years ago (73% vs. 32%), providing further evidence of an ongoing improvement in prognosis. Ó 2006 Elsevier Ltd. All rights reserved. 1. Introduction Since the early 1970s there have been impressive and contin- uing improvements in prognosis for childhood tumours. 1–7 Both short-term 8,9 and long-term 10–13 mortality appear to have reduced, although the latter is still not negligible for some types of childhood cancer. 12 Long-term survival remains the most direct measure of outcome for children with cancer. However, long-term sur- vival is usually estimated from the observed experience of co- horts of patients diagnosed many years in the past. For example, 25-year survival has typically been estimated from the survival experience of patients diagnosed 25 years ago and followed-up until the current day. When there are contin- uing improvements in survival, such historical data will not provide valid predictions of survival of patients diagnosed currently who will experience the benefits of recent improve- ments in therapy. It is, therefore, important to use more valid 0959-8049/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejca.2005.08.046 * Corresponding author: Tel.: +44 117 9287296; fax: +44 117 9287325. E-mail address: l.zuccolo@bristol.ac.uk (L. Zuccolo). EUROPEAN JOURNAL OF CANCER 42 (2006) 1135 – 1142 available at www.sciencedirect.com journal homepage: www.ejconline.com