You need JavaScript to view this

The selection and use of control groups in epidemiologic studies of radiation and cancer

Abstract

Current risk estimates for radiation-induced cancer are based on epidemiologic studies of humans exposed to high doses of radiation. A critical feature of such studies is the selection of an appropriate control group. This report presents a detailed examination of the principles underlying the selection and use of control groups in such epidemiologic studies. It is concluded that the cohort study is the preferred design, because of the rarity of exposure to high levels of radiation in the general population and because the cohort design is less susceptible to bias. This report also assesses potential bias in current risk estimates for radiation-induced cancer due to inappropriate choice and use of control groups. Detailed summaries are presented for those epidemiologic studies on which the BEIR IV risk estimates are based. It is concluded that confounding is by far the major potential concern. Bias is probably negligible in risk estimates for breast cancer. For lung cancer, risk estimates may be underestimated by about 30 percent for males and 10 percent for females due to confounding of smoking and radiation exposure. For leukemia and cancers of the thyroid and bone, the absence of established non-radiation risk factors with a high prevalence in the  More>>
Publication Date:
Sep 01, 1990
Product Type:
Technical Report
Report Number:
INFO-0376
Reference Number:
SCA: 560151; PA: AIX-24:005915; SN: 93000931035
Resource Relation:
Other Information: PBD: Sep 1990
Subject:
63 RADIATION, THERMAL, AND OTHER ENVIRON. POLLUTANT EFFECTS ON LIVING ORGS. AND BIOL. MAT.; CARCINOGENESIS; EPIDEMIOLOGY; DELAYED RADIATION EFFECTS; ERRORS; HUMAN POPULATIONS; MORTALITY; RISK ASSESSMENT; 560151; MAN
OSTI ID:
10118807
Research Organizations:
Atomic Energy Control Board, Ottawa, ON (Canada)
Country of Origin:
Canada
Language:
English
Other Identifying Numbers:
Other: ON: DE93611533; TRN: CA9200961005915
Availability:
OSTI; NTIS (US Sales Only); INIS
Submitting Site:
INIS
Size:
[187] p.
Announcement Date:
Jun 30, 2005

Citation Formats

Howe, G R, Friedenreich, C M, and Howe, P D. The selection and use of control groups in epidemiologic studies of radiation and cancer. Canada: N. p., 1990. Web.
Howe, G R, Friedenreich, C M, & Howe, P D. The selection and use of control groups in epidemiologic studies of radiation and cancer. Canada.
Howe, G R, Friedenreich, C M, and Howe, P D. 1990. "The selection and use of control groups in epidemiologic studies of radiation and cancer." Canada.
@misc{etde_10118807,
title = {The selection and use of control groups in epidemiologic studies of radiation and cancer}
author = {Howe, G R, Friedenreich, C M, and Howe, P D}
abstractNote = {Current risk estimates for radiation-induced cancer are based on epidemiologic studies of humans exposed to high doses of radiation. A critical feature of such studies is the selection of an appropriate control group. This report presents a detailed examination of the principles underlying the selection and use of control groups in such epidemiologic studies. It is concluded that the cohort study is the preferred design, because of the rarity of exposure to high levels of radiation in the general population and because the cohort design is less susceptible to bias. This report also assesses potential bias in current risk estimates for radiation-induced cancer due to inappropriate choice and use of control groups. Detailed summaries are presented for those epidemiologic studies on which the BEIR IV risk estimates are based. It is concluded that confounding is by far the major potential concern. Bias is probably negligible in risk estimates for breast cancer. For lung cancer, risk estimates may be underestimated by about 30 percent for males and 10 percent for females due to confounding of smoking and radiation exposure. For leukemia and cancers of the thyroid and bone, the absence of established non-radiation risk factors with a high prevalence in the population under study suggests that there is unlikely to be any substantial confounding radiation risk estimates. Finally, lifetime excess mortality risks have been estimated for several of the cancers of interest following exposure to radiation based on Canadian age-, sex- and cause-specific mortality rates. It is concluded that errors in measurement exposure, uncertainty in extrapolating the results of high dose studies to low doses and low dose rates, and sampling variation in the epidemiologic studies contribute far more to uncertainty in current risk estimates than do any biases in the epidemiologic studies introduced by inappropriate selection and use of control groups. (161 refs., 19 tabs.).}
place = {Canada}
year = {1990}
month = {Sep}
}