Measuring childhood cancer late effects: evidence of a healthy survivor effect
Given considerable focus on health outcomes among childhood cancer survivors, we aimed to explore whether survivor bias is apparent during long-term follow-up of childhood cancer survivors.
We identified all 1-year survivors of cancer diagnosed before 20 years of age in Denmark, Finland, Iceland, and Sweden. From the general population, we randomly sampled a comparison cohort. Study individuals were followed for hospitalizations for diseases of the gastroenterological tract, endocrine system, cardiovascular system, or urinary tract from the start of the cancer registries to 2010. We estimated cumulative incidence with death as competing risk and used threshold regression to compare the hazards of the diseases of interest at ages 20, 40, 60, and 75 years.
Our study included 27,007 one-year survivors of childhood cancer and 165,620 individuals from the general population. The cumulative incidence of all four outcomes was higher for childhood cancer survivors during early adulthood, but for three outcomes, the cumulative incidence was higher for the general population after age 55 years. The hazard ratios (HRs) decreased for all outcomes with increasing age, and for two of the outcomes, the hazards were higher for the general population at older ages (endocrine diseases: age-specific HRs = 3.0, 1.4, 1.0, 0.87; Cardiovascular diseases: age-specific HRs = 4.1, 1.4, 0.97, 0.84).
Our findings provide empirical evidence that survivor bias attenuates measures of association when comparing survivors with the general population. The design and analysis of studies among childhood cancer survivors, particularly as this population attains older ages, should account for survivor bias to avoid misinterpreting estimates of disease burden.
KeywordsSurvivor bias Healthy survivor effect Cancer survivorship Childhood cancer Competing risk Late effects
We thank Andrea Bautz and Anja Krøyer Kristoffersen for their for tremendous contribution to data management and members of the ALiCCS board Catherine Rechnitzer, Kirsi Jahnukainen, Jørgen H. Olsen, Finn Wesenberg, and Lars Hjorth for their valuable guidance and discussions. The study was supported by Grant 09-066899 from the Danish Council for Strategic Research and by a grant from the Danish Child Cancer Foundation. All authors have approved the final manuscript. Peter H. Asdahl had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 1.Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd ed. Philadelphia, PA: Lippincott, Williams Wilkins; 2008.Google Scholar
- 19.Xiao T, Whitmore GA, He X, Lee M. Threshold regression for time-to-event analysis: the stthreg package. Stata J. 2012;12:257–83.Google Scholar
- 20.Coviello V, Boggess M. Cumulative incidence estimation in the presence of competing risks. Stata J. 2004;4:103–12.Google Scholar
- 21.Bhakta N, Liu Q, Yeo F, Baassiri M, Ehrhardt MJ, Srivastava DK, et al. Cumulative burden of cardiovascular morbidity in paediatric, adolescent, and young adult survivors of Hodgkin’s lymphoma: an analysis from the St Jude Lifetime Cohort Study. Lancet Oncol. 2016;17:1325–34.CrossRefPubMedPubMedCentralGoogle Scholar