Baseline survey.The objective for the preinterven-
tion household survey was the personal interview of
1600 women residing in the neighborhoods or communi-
ties served by the eight public health clinics in BACCIS.
The census tracts to be sampled were selected by plot-
ting the addresses of a random sample of clinic atten-
dees on a map. Average monthly rent from the census
was used as a proxy for low-income status of the tracts
where the most addresses were located. Census blocks
were randomly selected within the census tracts based
on a proportional probability sample and a sampling
fraction computed on the basis of the number of house-
holds in each block. One woman, aged 40–75 years, was selected at random from each randomly chosen household. Survey workers were required to make four visits and three phone call attempts to a household to set up an interview before considering it a “nonขresponse.”