Table 1 Summary evidence in terms of Bradford-Hill Criteria [14] for power frequency magnetic fields causing childhood leukaemia.

Strength of Association

Pooled studies suggest a statistically significant doubling of risk above 0.3-0.4 uT. The strength of association is considered to be weak and only a small proportion of cases are attributable to high exposure.

Statistically significant raised risks of leukaemia are observed with increasing exposure to ionising radiation. Risk estimates are extrapolated from epidemiological data at higher doses using a linear no-threshold exposure response model.

Consistency

The association is observed almost exclusively in childhood case-control studies.

The association is observed in two different situations: first, studies of Japanese atomic bomb survivors irradiated as children, and second, studies of childhood cancer and antenatal exposure of the foetus to diagnostic X-rays.

Specificity

The association seems to be restricted to leukaemia, athough other childhood cancers have been investigated less frequently and less rigorously.

Studies have demonstrated that a number of different cancers are associated with exposure to ionising radiation.

Temporality

In ALL, the most common type of childhood leukaemia, the disease occurs relatively rapidly after exposure, normally in the third or fourth year of life.

In many of the cancers associated with ionising radiation, exposures can precede lesions by as much as several decades.

Dose response relationship

There are too few data, even after pooling, to identify the shape of a possible dose-response relationship. Threshold exposure response models have been suggested although data are also compatible with other trends.

A linear-quadratic dose response relationship is found between childhood leukaemia and ionising radiation exposure in A-bomb survivor studies, except at the highest levels of exposure. The shape of the dose-response curve is uncertain at low doses.

Biological plausibility

A number of mechanisms have been proposed for the interaction of magnetic fields with the human body, but it is unclear how these might affect the processes that lead to disease, particularly at the low levels identified in the epidemiological investigations. In vitro and in vivo experiments fail to show a consistent effect that might explain the development of childhood leukaemia.

There is a good mechanistic basis for suggesting ionising radiation causes leukaemia, involving direct damage to DNA. There are also other processes that have the potential to modify the simple model. There is abundant in vitro and in vivo evidence to support the carcinogenic effect of ionising radiation.

Biological coherence

The cause of childhood leukaemia is complex and not well enough understood to make an assessment.

The observed associations are consistent with what is known about the carcinogenic effects of ionisation radiation.

Experiment (reversibility)

Evidence that removing the exposure reduces disease would be difficult to ascertain because of the small fraction affected.

Evidence is difficult to ascertain.

Analogy

No analogies in adjacent parts of the electromagnetic spectrum.

A leukaemogenic effect is consistent with what is known about ionising radiation causing a range of cancers.