A simple explanation of a common concept in epidemiology
In our introduction to epidemiology we explain that just because you may discover that exposure A is associated with disease B, that doesn't necessarily mean that A causes B. The association could be a chance finding, it could be because of bias, or it could be caused by a confounding factor. We explain confounding factors in more detail here.
Let's give an example - a rather extreme example, to make the point. If we did a study, we would probably discover that having yellow stained fingers is associated with lung cancer. That's not because stained fingers cause lung cancer. Fairly obviously, it's because both are caused by smoking. Smoking is the "confounding factor", which makes it look as if stained fingers may be causing lung cancer, when there is in fact no causal link.
Or another example: we might well find that tonic water is associated with cirrhosis of the liver. But tonic water doesn't cause liver disease. It's alcohol (specifically gin) that causes the disease, and because drinking tonic water is associated with drinking gin, it creates an association between tonic water and liver disease. Alcohol is the confounding factor.
Application to EMFs
Epidemiological studies show a statistical association between exposure to magnetic fields and childhood leukaemia. Is that causal - magnetic fields cause leukaemia - or is it confounding - something else, the confounding factor, causes leukaemia, but happens to be associated with magnetic fields?
The answer is that we don't know. On the one hand, exposure to magnetic fields is associated with many other factors - socioeconomic status, size of home, etc - see a longer listing. If one of those factors causes childhood leukaemia, it could be the confounding factor that explains the association with magnetic fields.
On the other hand, epidemiologists always try to adjust for possible confounding factors (almost all studies of EMFs adjust as a minimum for socioeconomic status) and it rarely makes much difference. Certainly, no-one has yet identified a strong candidate for confounding as the explanation. But the possibility remains open.
See also:
Other aspects of epidemiology: