Study sites were urban and suburban neighbourhoods within the
city of Shanghai and urban neighbourhoods and rural communities
within the official city boundaries of Tianjin. Because it is
possible that city residents are more likely to adopt communitybased
innovations, community randomization was stratified to
include urban, suburban and rural residents. The neighbourhoods
selected were those whose leaders were willing to collaborate
with the two schools of nursing. In China, access to
neighbourhoods is still through leaders designated by the local
authorities. Once the available urban and non-urban neighbourhoods
were identified in both Shanghai and Tianjin, they were
randomly assigned to treatment or control using random
number generation, so that for each type of area there were both
treatment and control communities. This type of cluster randomization
was used to minimize treatment contamination as
numerous women would be living in the same apartment buildings,
or otherwise, in close proximity