Mortality among 5-year survivors of cancer diagnosed during childhood or adolescence in British Columbia, Canada.

*INTRODUCTIONAdvances in cancer therapy over the last 40 years havedramatically improved survival for children and adolescentsdiagnosed with cancer [1–3]. Notwithstanding theseimprovements, there appears to be an increased risk oflonger-term mortality among survivors [1,4–7]. Previousinvestigations report increased risk of death due to diseaserecurrence (the major cause), occurrence of a new secondcancer, or other cause, and risk of later death appears to beinfluenced by factors such as gender, age at first diagnosis,and therapy [1,2]. Relative risk of later mortality amongchildhood and adolescent cancer survivors is reported torange between 10 and 20 times higher than that expected inthe general population [4–8]. However, there is limitedinformation on risk of death due to causes other than cancer,or the effect of more recent therapies. Few studies haveexamined very long term mortality or mortality in morerecent cohorts, and some were too small to examine cause-specific mortality [2]. Many institution-based studies havereported on cohorts with a selected group of patientscompared to the total population of survivors, or had limitedfollow-up [2,4,6].The aim of the present study is to assess the risk of overallandcause-specificlong-term mortality inapopulation-basedcohort of 5-year survivors of childhood and adolescentcancer diagnosed from 1970 to 1995 and followed to end2000 in British Columbia (BC), Canada. We also assessedthe effect of diagnosis, and demographic and temporalfactors on risk.METHODSIdentification of the Survivor CohortThe study cohort was identified from the BC CancerRegistry, and consisted of all BC residents diagnosed beforeage 20 years between 1 January 1970 and 31 December 1995with a primary cancer or tumor listed in the InternationalClassification of Childhood Cancers (ICCC) [9], and whosurvived at least 5 years from the time of their diagnosis.Demographic and diagnosis information were obtained,including full name (for linkage purposes only), birth date,gender, information on date, and cause of death (ifapplicable), and for each new primary diagnosis, date ofdiagnosis,andmorphologyandhistologycoded accordingto