We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Cancer epidemiology in practice: Working notes on cancer history-based selection and censoring.
- Authors
Berge, Leon A. M.; Grimsrud, Tom K.; Babigumira, Ronnie; Støer, Nathalie C.; Shala, Nita K.; Veierød, Marit B.; Stenehjem, Jo S.
- Abstract
Researchers conducting observational studies of exposure-cancer associations must make decisions regarding the selection and censoring of study subjects with respect to their cancer history. Given available information, this may include cancers prior to start of follow-up (prevalent cases), and multiple primary cancers (MPCs) at the same or different time-points during follow-up. The choice of analytical strategy may have implications for statistical power and for potential biases. We discuss herein two approaches, the "all cancer approach" and the "first primary cancer approach". Each contains a set of criteria, primarily for, but not limited to, cohort studies of exposure-cancer associations. Both approaches exclude subjects with prevalent cancers at start of follow-up to avoid information and selection bias but differ in terms of case definition and censoring criteria. In the "all cancer approach", subjects are censored on non-eligible cancer subtypes of the cancer under study, though subjects with MPCs are not censored on other incident cancers preceding or occurring simultaneously (at time of diagnosis) with the cancer under study. To limit the influence of prior treatment regimens, the "first primary cancer approach" censors subjects upon any first primary incident cancer not under study, including any "simultaneous" MPCs, as well as non-eligible subtypes of the cancer under study. Depending on the cancer under study, these approaches may yield different estimates of the exposure-cancer association and may in this regard be utilized for sensitivity analyses. These alternative approaches are presented and discussed here as a potential resource and source of further discussion for peers in the field of cancer epidemiology.
- Subjects
PREGNANCY; BREAST cancer; RENAL cancer; CANCER diagnosis; BOLEN test
- Publication
Norsk Epidemiologi, 2022, Vol 30, Issue 1/2, p107
- ISSN
0803-2491
- Publication type
Article
- DOI
10.5324/nje.v30i1-2.4986