_ A~~.E N D IX _ B _
Experimental
Appendix B. Experimental Design of the China Study
SIXTY-FIVE COUNTIES in twenty-four different provinces (out of twenty-
seven) were selected for the survey. They represented the full range
of mortality rates for seven of the more common cancers. They also
provided broad geographic coverage and were within four hours' travel
time of a central laboratory. The survey counties represented:- semitropical coastal areas of southeast China;
- frigid wintry areas in northeast China, near Siberia;
- areas near the Great Gobi desert and the northern steppes;
- and areas near or in the Himalaya Mountains ranging from the far
northwest to the far southwest part of the country.
Except for suburban areas near Shanghai, most counties were located
in rural China where people lived in the same place their entire lives
and consumed locally produced food. Population densities varied wide-
ly, from 20,000 nomadic residents for the most remote county near the
Great Gobi desert, to 1.3 million people for the county on the outskirts
of Shanghai.
This survey is referred to as an ecological or correlation study design,
meaning that we are comparing diet, lifestyle and disease characteristics
of a number of sample populations, in this case the sixy-five counties.
We determine how these characteristics, as county averages, correlate
353