A review of the types of childhood cancer associated with a medical X-ray examination of the pregnant mother

For 65 years the interpretation of the statistical association between the risk of cancer in a child and a prior diagnostic X-ray examination of the abdomen of the pregnant mother has been debated. The objections to a direct cause-and-effect explanation of the association vary in their strength, but one of the most notable grounds for controversy is the finding from the first and largest case-control study reporting the association, the Oxford Survey of Childhood Cancers (OSCC), of an almost uniformly raised relative risk (RR) for nearly all of the types of cancer that are most frequent in children. Here we compare the antenatal X-ray associations found in the OSCC for different types of childhood cancer with the results of all other case-control and case-cohort studies appropriately combined in meta-analyses, and we also review the findings of the few cohort studies that have been conducted.

From the case-control/case-cohort studies other than the OSCC there are consistent and clear elevations of risk for all types of childhood cancer combined, all leukemia, and all cancers except leukemia combined. This compatibility of the findings of the OSCC with those of the combined other studies is less clear, or effectively absent, when some categories containing smaller numbers of incident cases/deaths are considered, but lack of precision of risk estimates due to sparse data presents inferential challenges, although there is a consistent absence of an association for bone tumors. Further, more recent studies almost certainly address lower intrauterine doses, with an anticipated decrease in estimated risks, which could be misleading when comparisons involve a limited number of studies that are mainly from later years, and a similar problem arises when having to employ all types of antenatal X-ray exposures for a study because data for abdominal exposures are absent. The problem of low statistical power is greater for cohort studies, and this, together with other shortcomings identified in the studies, limits the interpretational value of results. The findings of non-medical intrauterine exposure studies are constrained by sparse data and make a limited contribution to an understanding of the association. Certain aspects of the various studies require a need for caution in interpretation, but overall, the appropriate combination of all case-control/case-cohort studies other than the OSCC lends support to the inference that low-level exposure to radiation in utero proportionally increases the risk of the typical cancers of childhood to around the same level.