We explain here that some mechanisms proposed to explain how power-frequency magnetic fields might affect human health actually depend on the total magnetic field - the instantaneous sum of the earth's static magnetic field and the AC field from the power system - not just on the AC field.  One example of a mechanism like this is the free radical mechanism.  We might expect that these mechanisms would produce a rather bigger effect from the tens of microteslas changes in the geomagnetic field over the earth's surface than from the sub-microtesla fields implicated in the epidemiology of childhood leukaemia. (Especially as the AC variation itself should largely average to zero over time.)

Ideally, we'd like to test this by looking directly for health effects of the geomagnetic field.  But this is problematical.  The geomagnetic field basically distinguishes the tropical lattitudes - low geomagnetic fields - from the temperate lattitudes - higher fields.  So many other factors vary in the same way it seems unlikely we could ever disentangle any effects of the geomagnetic field.

A second best is to look at whether the geomagnetic field affects the results of epidemiological studies of AC fields - in the jargon, is the geomagnetic field an effect modifier?  Is the effect of 0.4 µT AC superimposed on top of 57 µT (Canada, New Zealand) the same as 0.4 µT AC superimposed on top of 23 µT (Brazil)?

This has been tested in a scientific paper.  This is the graph from the paper showing the relative risks from the individual epidemiological studies of AC fields plotted against the geomagnetic field in those countries:

Unfortunately, this doesn't really tell us much.  There is a trend, but it's a weak one, partly because so many of the studies are bunched together with similar geomagnetic fields (these are the studies in Europe, north America and New Zealand).