A Methodological Issue in the Analysis of Second-Primary Cancer Incidence in Long-Term Survivors of Childhood Cancers

Yutaka Yasui, Yan Liu, Joseph P. Neglia, Debra L. Friedman, Smita Bhatia, Anna T. Meadows, Lisa R. Diller, Ann C. Mertens, John Whitton, Leslie L. Robison

Research output: Contribution to journalArticlepeer-review

65 Scopus citations


Survival of childhood cancer patients has increased remarkably in the last several decades due to therapeutic improvements. Associated with this progress is the emerging need to accurately assess/minimize late effects of cancer therapy in long-term survivors. This paper considers a methodological issue in assessing the risk of second-primary malignant neoplasms, a major late effect of concern, using second-primary female breast cancer as an example. In the assessment of second-primary malignant neoplasm risk, attained age is a critical factor that must be taken into account. Even with follow-up of decades, childhood-cancer survivors are still at relatively young ages for developing adult-onset diseases. Attained ages at follow-up, however, modify cancer risk considerably; for example, in the general population, women aged 40 years have about fivefold increased breast cancer risk compared with women aged 30 years. A failure to account for the natural age-associated increase of risk could alter, or even reverse, analytical conclusions. This problem was studied empirically by both descriptive and regression analyses of two major studies of long-term childhood-cancer survivors, the Childhood Cancer Survivor Study (1975-1999) and the Late Effects Study Group (1955-1994). These showed appreciable differences in the analytical results by not accounting for the natural age-associated increase of risk illustrating a significant impact of this methodological issue on study conclusions.

Original languageEnglish (US)
Pages (from-to)1108-1113
Number of pages6
JournalAmerican journal of epidemiology
Issue number11
StatePublished - Dec 1 2003

Bibliographical note

Funding Information:
This work was supported by grant U24-CA-55727 of the National Institutes of Health.


  • Age distribution
  • Age factors
  • Cohort studies
  • Epidemiologic methods
  • Incidence
  • Models
  • Neoplasms
  • Proportional hazards models
  • Second primary
  • Statistical

Fingerprint Dive into the research topics of 'A Methodological Issue in the Analysis of Second-Primary Cancer Incidence in Long-Term Survivors of Childhood Cancers'. Together they form a unique fingerprint.

Cite this