Commentary and Review Series 1-90
A nested case-control approach to interactions between radiation dose and other factors as causes of cancer
Land CESummary
Often a nested case-control study is the most practicable approach to estimating the interaction of two cancer risk factors in a large cohort. If one of the factors has already been evaluated for the entire cohort, however, more information is already available about its relationship to risk than could be obtained from a nested study. A modified case-control approach is proposed, in which information about the second, unknown factor is sought for cases and controls matched on the first factor. The approach requires, for interaction models other than the multiplicative, a nonstandard analytical approach incorporating cohort-based information about the first factor. The problem is discussed in the context of breast cancer risk in a defined cohort of female Japanese atomic bomb survivors, in relation to radiation dose and reproductive history.
Often a nested case-control study is the most practicable approach to estimating the interaction of two cancer risk factors in a large cohort. If one of the factors has already been evaluated for the entire cohort, however, more information is already available about its relationship to risk than could be obtained from a nested study. A modified case-control approach is proposed, in which information about the second, unknown factor is sought for cases and controls matched on the first factor. The approach requires, for interaction models other than the multiplicative, a nonstandard analytical approach incorporating cohort-based information about the first factor. The problem is discussed in the context of breast cancer risk in a defined cohort of female Japanese atomic bomb survivors, in relation to radiation dose and reproductive history.